I

I start with the assumption that statutes that are directives must be capable of guiding the conduct of those they address. Shapiro has distinguished two kinds of guidance he finds in the writings of Hart: motivational and epistemic.⁶

This article is neutral on the question of whether statutes that are directives must be capable of motivational guidance. But it is not neutral with regard to epistemic guidance. Statutes that are directives must be capable of epistemic guidance. They must be capable of providing information—about actions required, prohibited, or permitted—that people can act upon when they are so motivated. This capacity is one of the reasons why it is rational to create such directives in the first place and one of the reasons why it is rational to promulgate them.

How should the idea of epistemic guidance be understood? Shapiro puts it this way: “A person is epistemically guided by a legal rule when he learns of his legal obligations from the rule provided by those in authority and conforms to the rule [on the basis of that knowledge].”⁷

Clearly, this description applies only to directives that require or prohibit conduct. But one might be epistemically guided by legal permissions as well, if those permissions make available a type of conduct that would not be possible to engage in but for the law—for example, the exercise of legal powers. (One is epistemically guided by a wills statute, for example, when one follows the steps prescribed in the statute in order to exercise a power the statute makes possible.)

There is another reason why Shapiro's account of epistemic guidance should be treated as a first approximation. What is needed is an account of how statutes epistemically guide, rather than how legal rules do. It is a matter of law what the legal rules are, and, as the previous discussion noted, it appears that what the legal rules require may diverge from what the statutes require. The points made so far suggest that Shapiro's account of epistemic guidance should be recast as follows: a person is epistemically guided by a statute when he learns of putative obligations or permissions from the statute and conforms to it on the basis of that knowledge.

Sharpening the idea of learning about putative obligations will be helpful. Although epistemic guidance may be discussed as if knowing the rule the statute sets down constitutes all the information necessary for epistemic guidance, such understanding must embrace act-tokens. A person must be able to identify appropriate act-tokens picked out by the rule in circumstances in which it is picked out in order for that person to be epistemically guided. (There may be indirect epistemic guidance, as when a statute prohibits specific sorts of acts and an overly cautious person refrains from doing any acts of a more general type. In this case, the person is following a broader directive that picks out a larger class of acts than does the statute. This may occur when a prohibitory statute has a chilling effect, and especially if the statute as a whole is vague. I will ignore this complication in what follows, as this type of guidance is both atypical and, from the point of view of ordinary legislation, pathological. Moreover, it requires the same ability: the ability to identify act-tokens of the new, invented—albeit not canonically stated or authoritatively given—directive, some of which the individual, to be epistemically guided, must know are also picked out by the statutory provision.)

There are various other conditions for statutes to guide epistemically, such as that they are made public rather than hidden, that they actually require or prohibit (or permit) particular act-tokens, that these acts are not impossible for their addressees, and so on. Some of Fuller's conditions of the Internal Morality of Law are applicable here. Another condition is that if the statute is promulgated, it is promulgated in language in which its addressees are competent. But are there still other conditions of a very different sort that must be added to ensure that addressees can correctly apply the statute to act-tokens? There are, according to intentionalism.

Intentionalism is the view that legislative intent determines a statute's application-meaning. “Intentionalism” is not meant to denote any of the many varieties of the view that legislative intentions have some proper role in applying statutes when the application of those statutes is otherwise unclear or indeterminate. I have no quarrel with a view that recommends recourse to legislative intentions, where they exist and can be known, in some cases, and even in cases that may be morally and legally significant. Indeed, I have no quarrel with the view that recommends recourse to legislative intentions, where they exist and can be known, to modify statutes when applying the statutes would have manifestly absurd results.

Intentionalism as I use the term is a claim about all cases in which a statute requires (or prohibits or permits) acts. The intentionalist claim is not a normative one, to the effect that officials ought to find applications by consulting legislative intentions. The claim is a conceptual one about application-meaning: the meaning of a statute is determined by legislative intentions relevant to its application.

There are weaker and stronger conceptions of legislative intention. On the weakest version, legislative intention is entailed by—instead of determinative of—the application-meaning of the statute. Necessarily the legislature intends however it is that the statute applies. Suppose, for instance, that the statute applies in accordance with its plain meaning. Then the average addressee could look at Hart's well-known “vehicles are prohibited in the park” statute and reason that since driving a city bus full of commuters into the park is plainly prohibited by the statute, it follows that it is the legislative intention to prohibit this activity. Such is the legislative intention whether or not legislators would have agreed, thought so at any point, intended to aim at a purpose served by barring commuter buses, and so on.

Application-meaning is determined, on this view, without recourse to anything like an investigation of how legislators thought or hoped the statute applied or what they may have meant by the statutory language. Obviously, a claim that application-meaning is determined by legislative intention, when legislative intention is understood in this way, dissolves into the claim that something else determines application-meaning. For that reason, intentionalism must be understood to use a different idea of legislative intent.

Legislative intention can instead be understood as entailed by the reasonable purpose of a statute, or the purpose that best fits the statutory plain meaning. Then accounts of application-meaning will draw some relation between the application of the statute and the purpose. For instance, it might be held that a statute really requires an act-token if the act of that type under the specified type of circumstances would, if generally performed, tend to serve the purpose. This sort of account thus dissolves the idea that legislative intention determines statutory application-meaning into the view that purpose determines statutory application-meaning. This latter view has been well criticized in the literature, and there is no need to repeat the discussion here.⁸

For this reason, the discussion of intentionalism will not be concerned with this version either.

One might hold instead that there is an independent kind of intention that legislatures creating statutes have concerning applications, and that this kind of intention determines the applications of statutes. This is the theory labeled here “intentionalism.” It has varieties, depending on what kind of independent intention is meant. One variety latches onto a purpose intention that legislatures are asserted to have with regard to statutes. (The purpose intention needn't be a reasonable purpose or one that best fits the statutory language, thus distinguishing this view from the previous one.) For instance, by Hart's “vehicles are prohibited in the park” statute, legislators may have intended to create a place of pedestrian safety in the park. Another variety stresses specific application-intentions that legislatures are alleged to have with regard to statutes—for example, by the same statute, legislators may have intended to exclude cars driven by commuters and delivery trucks on commercial errands. In any case, the legislative intention that counts is logically independent of plain meaning (understood, for the time being, as the analogue of sense for directives); though if the statute has been drafted with care, the intention explains why the legislators chose to express the statute in terms with that plain meaning. It is this independent intention that determines which act-tokens people are required to do, prohibited from doing, or permitted to do by a particular statute. (It is probably unnecessary to add that on this view, the legislative intention is thicker than the thin intention to enact a particular statute.) Still other varieties of intentionalism postulate more complex intentions.

Now, the view that the analogue of reference for a statute is entirely determined by independent legislative intentions seems wildly counterintuitive. Imagine a legal system with a single legislator. Imagine that he has created a statute that reads, “It shall be unlawful for any person to deface any signboard, signpost, lamp or lamppost, or any other property of the state.” Finally, imagine that the legislator, being a particularly strange fellow, intends by this very statute to prohibit dog-walking in a certain public park on Sundays. Just as there is a difference between what his statute means and what he means by it (a point that can be acknowledged by someone who says that the senses are different), there is a difference between how his statute applies and how he intends it to apply. The proposed view conflates the latter two in such a way as to absurdly eliminate the constraints of the linguistic component of a statutory directive. Moreover, it seems utterly impossible to reconcile this extreme view with the fact that the language of statutes is canonical.

In response, the intentionalist may deny a crucial premise of this thought experiment. He may maintain that it is incoherent or logically impossible to intend to prohibit dog-walking by issuing a statute in the suggested terms—at least, when the statute is written in the bill and the bill is read by and to the legislature that passes it. Nothing, she may claim, will count under these circumstances as the legislature having that particular intention. This response is doubtful. Recall that the legislative intention that counts is logically independent of plain meaning. It would seem to follow that the former could vary widely from the latter. Although analogy with individual speakers is suspect, for reasons discussed later, it seems that just as individuals may wildly misspeak, so can legislatures.⁹

However, let us grant that there are limits on what can count as the relevant legislative intentions and further grant that some such limits are set by the plain meaning of the statute. The intentionalist can then say that there must be the right relationship between the plain meaning of the statute and the independent legislative intentions in order for the latter to determine the application of the statute. (Then, in the statutory example, it is the kinds of defacing the legislator intended to ban, or the circumstances under which he intended to ban it, or some purpose he intended to achieve by the banning defacing as ordinarily understood, that determines how the statute applies.) He can also say that this relationship might not be very tight, so that legislative intentions can correct plain meaning to some extent.

Admitting that plain meaning determines which independent intentions have nothing to do with the application of a statute may be tricky for the intentionalist, because he will have to explain what it is about plain meaning that limits plain application. He will want to be careful not to give an answer that implies that plain meaning has sufficient application-determining effect on its own. I am not sure he can come up with a successful answer, one that is not ad hoc. But let us waive these difficulties and assume that they can be surmounted.

There is already much in the legal and philosophical literature to challenge the plausibility and even the coherence of the claim that there are such determinate legislative intentions when legislatures are corporate entities. This is particularly true when the questions about such intentions are distinguished from questions about the coherence of collective intentional action or collective intentions to enact a particular statute.¹⁰

Rather than enter this debate, I will focus instead on the adequacy of the account that the intentionalist can give of the epistemic guidance of statutes and the consequences for the rationality of promulgation.

Recall that a person is epistemically guided by a statute when he learns what it is he is required or permitted to do or prohibited from doing from the statute and conforms to it on the basis of that knowledge. Previous discussion established that such a person must learn something that extends to act-tokens picked out as required, prohibited, or permitted. He must know them as required, prohibited, or permitted by the statute. Now, if intentionalism is true, this state of affairs requires several conditions to be met in addition to the promulgation of the statute. It will not suffice merely to have the addressees linguistically understand the statute to some extent, as we have already assumed that they do. It will not suffice merely for them to know to what act-tokens the statute seems to be applicable, even if this is something that would be obvious to all. (I refer now to what I will call the plain application of the statute.) In the case in which the legislative intention diverges from the plain application, the addressees must have additional knowledge. They must know what act-tokens, contrary to first impressions, the statute really picks out. Their first impressions, drawn from plain application, must be corrected, and it would seem that only additional information will do the correcting.

Matters are not noticeably simpler in the case in which the legislative intention coincides with plain meaning. That coincidence must take place, of course, so that is another condition of epistemic guidance, but the intentionalist cannot stop there. The addressees have a belief (based, I want to emphasize, on plain application) that is true. However, they do not yet know that the statute picks out the act-tokens they correctly believe to be picked out, and so they have not yet learned from the statute which act-tokens are picked out. It is by now a philosophical truism that true belief does not amount to knowledge. On the standard analysis of knowledge, it takes justified true belief to amount to knowledge. This justification must be sufficient for knowledge. (Some epistemologists refer to the justification required as complete justification.) And so far, our addressees have not enough justification to qualify as knowing about the act-tokens. They would need to rely on something else. For example, they might rely on the generalization that the legislature means what it says and intends the plain applications (unless, let us suppose, the addressees are notified otherwise). However, that generalization would have to be true, and there must be specific social conditions that make it true. Addressees cannot assume that it is true in every society under all conditions¹¹

or for all statutory directives. Even those philosophers who reject the standard analysis of knowledge hold that some other factor or factors must be present in order for a true belief to constitute knowledge. So there must be additional conditions for epistemic guidance, and the possibility of epistemic guidance depends upon all of these conditions.

In short, on any plausible analysis of knowledge, something more than competence with the language and familiarity with certain kinds of social context that will allow an individual to understand plain meaning (and plain application) are needed in order for a statute to guide epistemically. This is true not only when the legislative intention fails to coincide with the shared beliefs of the addressees but even when the intention does coincide. When legislative intentions diverge from plain meaning, the additional conditions generally will not be met without more for the statute's promulgators, addressees, or other person to do or learn or undergo.

Worse, it follows that intentionalism cannot be wholly reconciled with the widely held and plausible view that, in general, promulgation is rational as the legislature's communication device for statutes. In some sense of “said,” all that addressees are officially given is what the legislature said.¹²

Why this is a rational practice if the intentionalist is correct in all that he claims determines application-meaning remains, for all the reasons given, something of a mystery. In particular, promulgation would not be rational as an epistemic guidance device when addressees do not know and cannot be expected to learn what applications the legislature intends (i.e., the specific social conditions for supplying this information are defective or absent). Promulgation of the text only will not be wholly rational as the legislative communication tool in circumstances in which it would have been helpful had the legislature added some separate official indication of the relevant application-intentions (including an official indication that they intend the plain applications).¹³ Promulgation of the statutory text alone would clearly fall short of rationality when the independent legislative intentions are at odds with the plain applications. This divergence is always possible, and there is typically nothing in promulgation to mark the cases where it obtains from ones where it does not. It follows that in conditions under which addressees cannot otherwise tell the difference, promulgation would not be wholly rational even as a method of providing fair warning (where the goal might be merely correct belief, not knowledge).

When one combines the observations made about intentionalism's handling of divergence of legislative intention from plain meaning with the price intentionalism pays in the cases of convergence with plain meaning, it becomes important to look for alternatives. Compared to the view that plain application determines application-meaning, Ockham's razor (or something close to it) is offended by the conditions intentionalists add for epistemic guidance in the case of statutory directives. So why believe that intentionalism is the more plausible view?

II

Intentionalism seems plausible, I suggest, because of its apparent connection with legislative authority. Treating the meaning of statutes as set independently of legislative intention appears to some philosophers to be inconsistent with recognizing the due authority of the legislature (when it exists).¹⁴

This view is put in its clearest form by Alexander. It is worth quoting his remarks at length: Alexander's view seems most plausible if one analogizes from situations such as these: a parent issuing directives to a minor child, a master issuing directives to a servant. These enterprises are ones in which those subject to the alleged authority are expected and know themselves to be expected to follow directives as the authority intends. Indeed, they are enterprises in which it would not be amiss for the ones subject to authority to anticipate the authority's directives by doing what the authority wishes done but has not yet attempted to communicate. Disregarding what the alleged authorities in these cases intend by their directives seems to disregard or disrespect their authority.

However, the analogy is problematic. I follow Shapiro in holding that authority in legislatures is not personal but impersonal.¹⁶

Legislative authority, like legislative actions, is institutional. What counts as the legislature's actions is determined by constitutive and other rules (including institutional procedural rules). Perhaps legislative intentions, where they exist, will be determined by constitutive and other rules. So there appear to be several relevant disanalogies with the person-to-person case. Moreover, multimember legislatures have a heavily text-bound procedure, as Waldron points out with respect to democracies.¹⁷ This insistence on the text (of drafts, amendments, bills, and so on) to identify what is being argued about, amended, voted on, or adopted is at odds with the alleged appropriateness of the addressees anticipating uncommunicated (and a fortiori unwritten) legislative intentions. (So is the practice of promulgation of the text.)

Nor will reflections about the grounds of legitimate authority in legislatures support Alexander's argument. Consider, for example, the most influential account of legitimacy in authority, one supplied by Raz. Raz offers a single analysis of both private and public practical authority. According to Raz, authorities possess legitimate authority insofar as they help their subjects to act on reasons.¹⁸

Raz's Normal Justification thesis states that Raz recognizes five situations in which someone may be a practical authority:²⁰ when he has a special expertise or wisdom; when he is less subject to bias and weakness; when the cost of deliberation is too high for others; when he is in a better position to accomplish certain ends (e.g., coordinate behavior); and when he is, as Shapiro puts it, in a position to solve a prisoner's dilemma.²¹ According to Raz, each of the five situations may ground legislative authority, although some grounds will be more common in some societies than in others.

How should we understand the legislative determination in order to preserve the possibility that legislatures can have legitimate authority on these five grounds? For the last four grounds, it is clear that no understanding of a determination that gives prominence to legislative application intentions is compelled. If the legislative directive is to be understood in accordance with the plain meaning of the text, there are ample scenarios for legislatures issuing such things to aid those subject to bias or weakness or to provide useful guidelines to those for whom it is too costly to deliberate. Indeed, for the reasons given in the preceding section and elaborated in Section V, a legislative directive that need be understood only in accordance with its plain meaning (and plain application) is a more efficient vehicle for solving coordination problems or prisoner's dilemmas than one that must track legislative application intentions.

This efficiency is due to the mechanism that intentionalism is intended to supplant, namely, the ability of people to understand the plain meaning (and discern the plain application) carried by a legislative text. This mechanism is at work if there are cases in which an independent legislative intention does not exist but the directive has a plain meaning and plain application. For once a directive has both, it can guide conduct, with the possible result that coordination problems or prisoner's dilemmas are solved. This point presents the intentionalist with dismal alternatives. He can say either that such directives do not have proper applications or that sometimes the proper application of the statutory directive is entirely determined by the plain application after all. This second alternative is ad hoc, and the first is little better. If being in a superior position to solve coordination problems or prisoner's dilemmas suffices for legitimate authority, the legislature may have legitimate authority precisely because it can issue directives with plain meaning, whether or not these directives are accompanied by (what the intentionalist will claim are) the appropriate independent intentions. It follows that to ignore the plain meaning when the independent intentions exist and are incompatible is, ironically, to undermine a possible source of the legislature's legitimate authority.

I have saved the expertise ground of legitimate authority for last. It might be claimed that here intentionalism seems particularly plausible.²²

In the case of a private person who is an expert on something (a plumber giving advice, say, on how to avoid a disastrous leak next time), we go out of our way, if we are rational, to make sure we have understood not merely the plain meaning of the plumber's pronouncements but what he meant by them and any other intentions of his that are relevant (such as specific application intentions). If the legislature has legitimate authority because it has special expertise on a matter, the correct understanding of what it has determined would also seem to take into account certain of its independent intentions, assuming they exist. However, since expertise is not commonly the basis of legislative authority, an argument from expertise cannot be used to defend the general claim of intentionalism.

Imagine, then, that the intentionalist objection is limited to the case of the expert legislature. It should be noted that any simple analogical argument based upon the plumber example is suspect because legislative actions (and intentions?) are institutional (not to mention formal). Without relying heavily on a person-to-person analogy, the intentionalist objection is both sustained and significant only if important questions are begged. To see this, consider that the following line of reasoning, purportedly in support of intentionalism, will not do:

For this argument is invalid. The legislature may be an expert. But by hypothesis, its directives have practical authority over nonexperts only when the nonexperts do better following the directives as they apply. It is entirely possible that the expert legislature fails to adequately translate what it knows into directives that meet this description. So the first conclusion in this chain argument does not follow. Moreover, directives have practical authority over their addressees only when the addressees do better following the directives as they apply. Unfortunately for the intentionalist, nothing follows from the origin of the directives or the expertise of that origin about how the directives apply. How the directives apply is the question; and without begging the question, it does not follow that they apply as the legislature intends.

One wants to protest on behalf of the intentionalist: surely the legislature, being an expert and being in a position to issue directives that reflect its expertise (whether or not the legislature has actually done so), has (in general) practical authority over nonexperts. This may be granted, but the intentionalist will derive no consolation. For not enough follows. One consequence is that nonexperts would do better if the legislature issued directives or any other sort of guidance that accurately reflected its expertise. Another consequence seems to be that nonexperts ought to seek guidance in whatever form they can from the legislature, or at least ought not to ignore such guidance when it can be attained from whatever source (not necessarily limited to the directives actually issued). But this is a prescription about how guidance ought to be sought (and it can be turned into a prescription about how guidance ought to be offered) in order fully to (recognize and realize the) value (of) the practical authority of legislatures when they are experts. However, the issue is neither about the practical authority of the legislature nor about fully valuing it. The issue is about the application of directives, upon which, on Raz's theory of authority, the practical authority of the directives hinges. Again, unless one assumes the truth of intentionalism, one cannot derive an answer to the question of how statutory directives actually issued by legislatures apply from claims about what it would take to fully value the practical authority of legislatures.

Nothing much changes if, instead of Raz's Normal Justification thesis, one substitutes other accounts of the grounds of legitimate legislative authority. For instance, Waldron suggests that legislatures may also become legitimate authorities when they are in a position to determine what reasons objectively apply to a group of people who disagree on a matter.²³

Shapiro embraces the Arbitration Model of traditional liberal political theory,²⁴ according to which legitimacy of political authorities arises from the subject's acceptance in some manner of the process by which authorities arbitrate between subjects and thereby resolve disputes. One way in which the authorities do this is by issuing directives. But one cannot draw the intentionalist objection on either view without begging questions.

In sum, in the situations that can ground legislative authority, an account can be given that takes the directive to be the rule set by the plain meaning of the text and takes the application-meaning to be what I have been provisionally calling its “plain application.” In some of the situations recognized by Raz as sufficient for practical authority, that approach seems more promising than the one offered by the intentionalist. I conclude that even in the case of an expert legislature, the intentionalist fails to sustain a serious, authority-based objection against alternative theories of statutory application.

III

If intentionalism is the Scylla that must be avoided, the school of thought that supposes that realist semantics, along with syntactical rules, determines the application-meaning of statutes is the Charybdis. There are intimations of this applied semantic realism view in Moore and Brink;²⁵

and though I am not sure that either maintains this view precisely as I shall describe it, they go far enough down this path that it is worth examining.²⁶

According to applied semantic realism, the words in statutes (typically) pick out genuine kinds whose members belong to the same kind because they have properties in common. Applied semantic realism has a great deal to say about the reference of concept-words used in statutes. It does not, unfortunately, have much to say about the application-meaning of units larger than concept-words, namely whole statutes or whole provisions of statutes; and from this failure stems its seminal shortcoming.

Applied semantic realism relies on a general semantic theory, one with implications beyond statutory interpretation and the legal domain. It claims to build on the semantic work of Putnam and Kripke. In the 1970s, Kripke and Putnam produced separate works arguing that natural-kind terms do not depend for their referents on a set of associated descriptions, however conventional or widely accepted. The referents depend, they argued, on complex relations to the world.

Applied semantic realism broadens the Kripke–Putnam approach to the claim that the referent of classifying terms is typically determined in some way by the way the world is. (Of course, without more specification about what that “some way” is, this description could include intentionalism.²⁷

For intentionalism holds that the application-meaning of statutes is determined by facts about the world, namely that legislators have particular independent intentions. Those facts about the world do not enter into a realist semantics for statutes, except in the case of a statute that explicitly refers to the legislators and their intentions.) Applied semantic realism is usually touted as being more compatible than certain alternative theories with substantive disagreement about the application of a term to a case (or the properties of things to which that term refers) and with changes of understanding about the referents of words.²⁸ So applied semantic realism allegedly accounts for the ability of later legislatures to rescind and later populations to comply with an elderly statute, one passed so long ago that popular ideas about the properties associated with terms used in it have changed.

The theory discounts or ignores the kind of legislative intention central to intentionalism, as Brink and Bix point out.²⁹

If the legislators were in error about the boundaries or the nature of the kind to which they were referring, applying their words according to realist semantics might frustrate their purposes and ignore their specific intentions.

Does it follow that the applied semantic realist has opened himself to the same objection as the intentionalist: commitment to a complicated picture of epistemic guidance, which makes the rationality of promulgation of statutes a mystery? It would appear that on applied semantic realism, epistemic guidance requires more than the promulgation of a statute and ordinary competences on the part of its addressees. Epistemic guidance requires in addition that the act-tokens people take to be prohibited (or required or permitted) by a statute be members of the appropriate kind. However, the applied semantic realist has a possible reply. The things in the world to which terms refer are not entirely divorced from usage. Uses of a term to denote paradigmatic instances determine to what genuine kind the term refers, and presumably native speakers of the language have mastered such usages. A statute can epistemically guide (in some cases) because part of its application-meaning is known to those who speak the language and read the statute.³⁰

The success of this answer depends in part upon on how the idea of paradigms is understood.³¹

If paradigms are in principle revisable in the light of discoveries, then the applied semantic realist has not avoided complicating the picture of epistemic guidance after all. For epistemic guidance will require more than promulgation and the competence that identifies paradigms. It must also be the case that the paradigms in question are of the right kind and so would not be subject to revision no matter how much of the truth were to be discovered.

Nonetheless, semantic realism applied to statutes has a certain appeal. It unproblematically provides a reason for regarding the text as important. It clearly allows for the possibility of legislative fallibility; legislators may create statutory rules whose applications they—individually or collectively—understand badly or not at all.

On the other hand, applied semantic realism has real difficulties. An obvious difficulty is the challenge of making coherent and plausible the view that general terms in statutory directives typically name genuine kinds.³²

One may mitigate this difficulty by limiting claims about application-meaning to statutes featuring certain kinds of terms—ones that allegedly name genuine kinds. The favorite candidates seem to be natural-kind terms and moral terms. (So, for example, applied semantic realism has something to say about the application-meaning of a prohibition on cocaine use or discrimination but perhaps not about the application-meaning of a statute concerning contracts or lawyers.) However, such a stratagem has a price: it limits the theoretical power of applied semantic realism and makes it an element in a larger, more heterogeneous (and ad hoc?) theory of application-meaning.

Even setting aside these difficulties, applied semantic realism is seriously incomplete. Clearly, the application of statutes is not determined by some simple function of the referents of its concept words and syntactical rules.³³

Fuller famously made the point that knowledge of the constituent words does not suffice for knowledge of the meaning—as in the applicability—of a statute.³⁴ Michael Moore gives an example on point. Suppose there is a federal statute making it a crime to “obstruct or retard the passage of the mail, or any driver or carrier,” and a county sheriff arrests a murder suspect who happens to be a mail carrier while he is on his route. The sheriff's actions appear clearly within the extension of the relevant predicates, yet at the same time, it is questionable that the statute applies to the sheriff's action.³⁵ Notice that the issue is not whether the statute ought to be applied to the sheriff's action. The question instead is whether the statute does apply. It is arguable that it does not.³⁶

Another example is attributed to Pufendorf. Suppose there is a statute in seventeenth-century Bologna, where dueling has become a serious social problem, that prohibits “drawing blood in the streets.” A surgeon who aids a passerby in an emergency is prosecuted under it. Though the surgeon has clearly drawn blood in the streets, it is not plausible to hold that the statute applies to his case.³⁷

Finally, consider a Canadian bylaw that required all “drug shops” to be “closed at 10 P.M. on each and every day of the week.” Defendants closed their drug shop at 10 P.M. and then opened it a few minutes later. They claimed that they had complied with the bylaw. But it seems safe to say that they had not.³⁸

There are situations outside the law that may be analogous. Consider what happens when parents tell the baby-sitter Wittgenstein's famous, “Shew the children a game,”³⁹

and learn later, to their consternation, that the sitter has taught the children “gaming with dice.” They protest, “I did not tell you to teach them that kind of game.” One way, but by no means the only way, of interpreting their protest is that they are claiming that their directive did not apply to that kind of game. Or consider MacCallum's literalist assistant, who is told to “bring back all the ashtrays you can find” for an important business conference.⁴⁰ Suppose that the zealous assistant rips built-in ashtrays off the walls of the building and snatches ashtrays from persons using them. One can hold, very plausibly, that the assistant misunderstood the directive. He was not being told to do what he did. Finally, consider what my legalistic teenager did in response to the command, “Pick those clothes up off the floor!” She picked them up, held them in the air for a moment, and let them return to the floor. That was not what she was being directed to do, and she knew it.

Without more, an applied semantic realist cannot explain the application-meaning of the bylaw that drug stores must close at 10 P.M. The act this law directs is not graspable by the concept of drug stores, the concept of 10 P.M., and the concept of closing in some aggregation. One gets legalistic or literalistic misunderstandings that way. The act directed is not graspable by discovery of the real property or properties that acts labeled in the past as “closings” have had in common. The same can be said about the real properties of acts labeled “drug store closings.” There have undoubtedly been drug store closings that lasted for short periods of time—for example, when the sole proprietor ran an errand or left for lunch. The question is whether these closings and a fortiori their genuine properties are relevant at all to what the statute requires. The claim that they are not is not the consequence of a theory (let alone a discovery) about the real nature of drug store closings, or closings, or any other concept mentioned in the bylaw. Similar claims may be made about the concepts in the other examples listed above.

The applied semantic realist may try to escape from these conclusions by denying that the statutes in question apply in the ways I have indicated. He may assert, instead, that they apply in the literalist ways I have rejected but add that judges should interpret them in the ways I have suggested. He may, in other words, press on the distinction, suggested at the beginning of this article, between the way a statutory provision applies and the way it is to be interpreted. However, this move effectively denies the possibility of literalist misunderstanding of statutes, and therein lies the problem.

No theory of meaning should deny the possibility of literalist misunderstandings, simpliciter. Perhaps a semantic realist need not do so for directives outside the legal domain. The general realist semantics assumed does deny the possibility of literalist misunderstandings of utterance meaning. However, the semantics might allow misunderstandings of speaker meaning. In other words, the theory will hold that the application or referent of what I say is typically determined by the boundaries of general kinds named by the general terms I use; but the theory can and should hold that I may mean something different from what I say. Then the theory might hold that the literalist mistake is failure to understand that I do not mean what I say.

However, there is still no room for the possibility of literalist misunderstanding of statutes. Recall that the applied semantic realist claims that the application of statutes depends on the meaning of the general terms used in the statute (along with syntactical rules). It would be abandoning applied semantic realism to postulate that the application of at least some statutes depends upon a(n institutional?) speaker meaning that differs from the meaning of the utterance, the written statutory provision. So even if the semantic realist can offer some account of literalist mistakes outside the legal domain, statutory language on applied semantic realism would be an area of (natural) language that no one could take too literally (though they could misunderstand on other grounds and though judges could interpret statutes too literally). Some of the nuances and subtlety of directive language one finds even in an everyday written directive (such as a note by a tenant to the handyman or by one roommate to another) would not be present in the case of legislative statutes.

Unfortunately, the conclusion that statutes, unlike other written directives, cannot be taken too literally is implausible. Moreover, in spite of the differences between statutes and other directives mentioned in the introductory section of this article, no independent reason presents itself for postulating a difference between statutory language and other language in this respect (i.e., for treating statutes as more literal than other directives). The fact that statutes are the result of a formal process, or the product of an institution, or uttered in at-remove contexts does not suffice to ground such a difference. Neither does the fact that statutory language is canonical. The applied semantic realist might argue that the existence conditions for a determinate (institutional) speaker meaning for statutes that differs from utterance meaning are met rarely, if at all. However, even if the claim were true, it would not support the view that statutes cannot be taken too literally unless one assumes the narrow account of literalist misunderstanding imagined above. That account, it must be noted, is neither compelling nor convincing unless one begs the question and assumes the truth of applied semantic realism or its general semantic assumptions. To be precise, the view that a literalist mistake cannot be a mistake about utterance meaning was designed to aid a very general and thoroughgoing semantic realism and has little to recommend it otherwise.

I contend, in short, that there does not seem to be a plausible, non-question-begging way for the applied semantic realist to avoid the counterexamples about statutory application offered above.

It will not help to be told, as Wittgensteinians claim, that concepts of closings, or games, or pickings-up are family-resemblance concepts. That may be true, but it is the membership in the family that is in question here: Which members of the family are relevant in this context? Semantic realism does not attempt to and cannot answer that question. There must be something else, some other mechanism that determines which of the paradigms of closings, or games, or pickings-up are the appropriate paradigms for the directives in question. After this determination, there may be room for semantic realism, because then it may make sense to imagine a search for the real nature of these paradigms and look for other things in the world that have the same nature. On the other hand, there may not be conceptual room for applied semantic realism once the mechanism for identifying the relevant members of the family is discovered and applied.

The fundamental question, therefore, is about this alleged additional mechanism. Should that mechanism be tied tightly, for instance, to what people who read the statutory provision would unhesitatingly take the statute to direct or not to direct, and should their grouping diverge from how things are really grouped, then the real nature of the referents of the concept words would be largely irrelevant to statutory application-meaning.⁴¹

As with intentionalism, the danger here is the ad hoc. A mechanism consistent with applied semantic realism must result in a principled or motivated combination. Certain familiar choices bring additional problems. For example, one approach is to claim that evident purpose determines which members of a larger set are in the application set of the statutory provision.⁴²

The reason that a closing of a drug store for a few minutes is not directed by the Canadian bylaw, on this view, is that closing a drug store for a few minutes does nothing, and quite obviously does nothing, to serve the evident purpose of the statute. Indeed, such a quick closing may frustrate the purpose.

But this view requires that an evident purpose inconsistent with brief closings be determined on grounds independent of the act-tokens in the application-meaning, and it is not clear how this can be done. On the contrary, an evident purpose for a statutory provision seems to be determined in part by the act-tokens to which that provision applies. The intuitive idea is that an evident purpose is evident precisely because it bears some relation to the act-tokens that constitute compliance with the provision—for example, it is likely to be, obviously will be, or can reasonably be expected to be served by them (individually or collectively, entirely or in general, etc.). In short, if it is evident to you that the purpose of the bylaw is inconsistent with brief closings, you are presupposing that ten-minute closings—and seven-minute closings, and five-minute closings, and so on, ad infinitum—are not within the bylaw's application-meaning. It is the assumption that brief closings are not in the application set that constrains the choice of an evident purpose for the bylaw, not the other way around.

Suppose instead that the applied semantic realist holds that the statutory provisions in question are elliptical. In doing so, he appeals to what is usually known as pragmatics, in contrast to semantics. The argument will then be that the ellipses are implicit because we would have to regard the legislature as irrational without them. There are two initial problems with the appeal to ellipsis. The first is that the obstructing mail statute does not appear to be elliptical. What seems to be doing the work of making the arrest permissible is an understood exception to the stated rule; yet specific exception clauses cannot, even in principle, be given exhaustively. This suggests that a mechanism different from ellipsis must be brought in to explain the obstructing mail example.

The second is that it is not clear that legislatures cannot be irrational in just the way the semantic realist suggests they are not. However, suppose it is granted that there are implied ellipses in the examples above. The appeal to ellipsis only pushes the problem back a level. For what determines the content of the ellipses? Semantic realism will not answer this question, and once it is answered, the challenge is to give a principled reason why the answer should not be regarded as supplanting semantic realism across the board.

Intentionalism purports to provide a method of determining which members of a family-resemblance concept used in a statute are apposite. Moreover, it is a unitary theory; with a single apparatus, it attempts to explain intuitions about applications in all of the statutory examples above. But intentionalism, as we have seen, accounts for epistemic guidance at a price and limits the rationality of promulgating statutory texts. If a simpler account can be given of the capacity of statutory directives to guide epistemically, then Ockham's razor, or its near equivalent, weighs in to provide a reason to reject intentionalism in its favor. Section IV presents an alternative to intentionalism that provides a simpler account of epistemic guidance that is also more liberal for the rationality of promulgation and that avoids the missing-mechanism problem of applied semantic realism.

IV

The simpler account may be summarized as follows. Statutes can epistemically guide because speakers of a language have the ability to identify act-tokens of a directive they read or hear in its linguistic and social context. However, this ability is not limited to or necessarily based upon knowledge of the real natures of paradigms or memberships in genuine kinds. It exists because in addition to whatever other competences speakers of a language must have to know the meaning of a directive (competence in semantics and syntactical rules), speakers are participants in certain social practices I will call practices of salient array-identification. Salient array-identification operates on statements and directives in contexts, not on isolated concept-words. And such practices have the capacity to operate in complete ignorance of and in conflict with the way that groupings occur in the real world and the independent intentions of utterers of the texts in question.

The idea of salient arrays is a metaphor. We can think of words and phrases that denote as having arrays of extensions. Some of these are, to paraphrase Finnis,⁴³

central in standard contexts, the standard ways we use these words and phrases. Others are peripheral in these contexts. We might think of the conceptual space as a map that has a standard orientation but also may be reoriented. Sometimes there are central clusters; sometimes there are many clusters, and the array is many-centered. When cognitive psychologists who study the way people categorize try to describe similarity relations used in categorizing, they have sometimes talked in terms of such arrays. They have imagined objects or items in the extensions represented as points in multidimensional space. So, for instance, according to Tversky and Hutchinson, for the denoting word “fruit” and speakers in contemporary American culture, a central array contains apple, banana, orange, and peach. Toward the periphery in one direction and relatively isolated is olive; in another direction, toward the periphery and relatively isolated, is coconut. Then there is a cluster off to one side consisting of honeydew, grapefruit, and pineapple.⁴⁴

Whole directives, too, I am suggesting, have clusters of act-tokens that they are about. When a word or a phrase is used in a directive, we can think of the directive as having a map of arrays that may or may not bear a tight relation to the map of a constituent word or phrase in its standard orientation and standard contexts. The map for the directive could, for example, be the result of reorienting or reconfiguring the map for a constituent word or phrase used in a standard context. (It should be made clear that nothing in the model I am suggesting assumes that words and phrases have an orientation, and thus array-salience, independent of their context.) There are broad social practices that make some clusters in the map of act-tokens for the directive salient under certain conditions and others obviously peripheral. These array-setting practices make use of cues and components that are not narrowly linguistic. They may include knowledge of how people do things, what is expected, the usual, and the norm.

So, for instance, in directives about amusing children in mainstream English-speaking cultures, the cluster-concept word “game” has a salient array that most certainly does not include gaming with dice. Instead, gaming with dice, like strip poker, gets shunted off to a peripheral cluster in this sort of context. The sitter ought to have known that, and therein lies her mistake. The literalist assistant ought to have known the salient cluster or clusters for the phrase “bring back ashtrays” in ordinary directives about business meetings among smokers and ought to have known that his kind of bringing back was far from that cluster. My legalistic daughter knew perfectly well that picking up clothing momentarily and then releasing it was nowhere near the salient cluster of the phrase “pick up … clothing” in the directive that I uttered, which is probably why she chose to do it. All three addressees in these examples exhibit an ignorance—or sham-ignorance—of broad, fundamental, social practices that directly bear on full linguistic competence. Indeed, the practices are so fundamental and the ignorance so unusual as to make these addressees appear stupid.

Whatever should be said, ultimately, about directives in all conversational contexts, my claim is that for statutory directives in the simplest cases, the application-meaning is given by the socially salient array for that directive—not because in the simplest case legislative intention coincides, but regardless of whether legislative intention coincides. When the federal statute makes it a crime to “obstruct or retard the passage of the mail, or any driver or carrier,” what is not in its salient array or arrays is a lawful arrest of a mail-carrying murder suspect, and “everybody knows it.” Shutting down a drug store and then reopening minutes later is nowhere near the salient array for the Canadian bylaw, as even the defendants must have recognized. (Consequently, it will surprise no one to learn that the court in R. v. Liggetts-Finlay dismissed the defendants' contention that they had complied with the bylaw.)

The theory proposed here, social salience theory, is not a conventionalist theory of reference. Social salience theory is not a general theory of reference at all. One of the reasons why it is not a conventionalist theory of statutory reference is that social salience need not be founded exclusively or even in part on convention (save the role conventions of language play in semantics and syntax). People in a society may find some applications of particular directives salient because of natural tendencies of mind and perception; they may find others salient because of a combination of natural tendencies and their internalization of background social conventions. Still, there is obviously a large role for convention to play in social salience, and it may be entirely because of socialization that people in a given society tend to converge on the same applications of a particular directive.

A persistent charge against conventionalist theories of meaning in law is that they cannot adequately represent deep disagreement about all aspects of law.⁴⁵

This charge is not, it seems to me, particularly damaging to social salience theory. Since it is not a general semantic theory, social salience theory does not render unintelligible what people are disagreeing about when they disagree about a simple case of statutory application. (Presumably they are each referring to the same directive and making claims about different act-tokens or arrays of act-tokens to which it applies.) Social salience theory concerns the truth conditions of the disputants' claims, not (other aspects of) the meaning of what they say when they make or deny statutory-application claims.⁴⁶

There is also a practical point to some argumentation in simple cases. One can sometimes trigger in another the recognition of the salience of an application by pointing out features of act-tokens (not necessarily their real natures or membership in genuine kinds) that he overlooks or misperceives. Then, too, what is salient may not be salient upon immediate inspection or thought. Salience may require reflection and even reasoning. In an earlier article, I used an example of the completion of an analogy on a practice Scholastic Assessment Test (S.A.T.) to illustrate this point.⁴⁷

But are there not cases of the application of a statute that are not simple and that permit of deeper or more substantive argument? Stavropoulos observes, on behalf of those frustrated with explaining away such disagreement, that “disputing lawyers perceive themselves to be arguing about whether cases in fact fall under the relevant legal concepts.”⁴⁸

This way of putting the matter is inadvisable, for reasons given in the last section.⁴⁹ But Stavropoulos's objection can be recast this way: disputing lawyers perceive themselves to be arguing about whether particular act-tokens in fact (or in truth) are picked out by a particular statute. If there are disputing lawyers, chances are that the application of the statute is not simple. Suppose there is lively debate about whether the Canadian bylaw permits a drug store to close its doors to business at 10 P.M. but reopen sharply at 12:01 A.M. Suppose that the issue in dispute cannot be resolved even in principle by getting one of the parties to see what is socially salient. Must social salience theory declare that necessarily the one who claims that the act-token is really prohibited by the statute makes a false claim? (Notice that his claim will not be incoherent according to social salience theory.) And if the claim is false, what are we to make of the argumentation that accompanies such claims?

Having suggested distinctions among claims at the beginning of this essay, I can postulate that the possibility of truth and the role of arguments vary with the type of claim made by the disputants. If the claim that the statute applies is one of pure application, it might be true, provided: (1) the instance is at the borderline of a salient cluster, or the borderline is itself vague; and (2) bivalence is preserved. (These conditions may not be compatible.) The claim that the statute applies could be the result of interpreting the statute, by which I mean assigning a meaning beyond the plain meaning (and its plain application). Such an assignment may be the result of a deciding whether to regard the provision as about closing for a conventional twenty-four-hour day or closing for the night until some usual time to resume business the next day. Perhaps there are constraints on acceptable interpretation such that the same decision on the instance in question is generated by all acceptable interpretations. Then one of the disputants can be correct, but social salience theory can maintain that it is correct about a subject other than the one social salience theory addresses.

The bylaw may already have been authoritatively interpreted in a certain way. If authoritative interpretations can alter the application a statute would have had in their absence, then social salience theory can admit that a claim in accordance with the authoritative interpretation can be true. If the question is about what the law prohibits or requires, the claim of a disputant could again be true or represent the right answer, but I have already given reasons to think it is an answer to a question not addressed in this essay. Finally, the claim of a disputant could be true or at least apposite about a rule associated with the statute if the statute occurs in a system with precedent and there have been judicial applications extending beyond or conflicting with (what would otherwise be) salient arrays, for reasons I detailed in an earlier work.⁵⁰

(This may happen when the statute is interpreted by judges but it may also happen when judges get the application wrong and their decisions have precedential weight.) So there is room in social salience theory to say that claims about a particular closing not in the salient array are not necessarily false; moreover, depending upon what constitutes acceptable interpretation and how a claim about the law is justified, there may be room to give reasons, and substantive ones at that.

Even when there is no antecedent fact of the matter about whether or not the kind of closing envisaged is prohibited by the bylaw, there is a point to maintaining, for example, that it is and trying to get others to view it that way. There can be a kind of coordinated cooperative activity among speakers in a community to whom arrays are socially salient. That same activity may yield the socially salient arrays originally. People in a sociolinguistic community have—or it would be rational for them to have—the generalized intention of coordinating the perception of groupings of referents of directives in order to make directive-issuing practices epistemically efficient; and also, when coordination is achieved between legal officials and laypersons, to maximize fair notice, improve predictability of official response, and so on.⁵¹

The process of working out which act-tokens are obviously directed by a statutory provision and which ought to be viewed as so directed can be a messy business in which arguments feature.⁵² It can also be a dynamic business in which the boundaries of socially salient arrays change over time and in response to the effects of persuasive arguments, enlarging to encompass what is nearby in the socially defined conceptual space or altering radically altogether. This change can itself be for reasons, reasons that justify and may be emotionally and practically compelling.

Someone who is willing to grant that directives apply to socially salient arrays may balk at the idea of extension through social processes and opt for something conceived of as more objective. On this view, statutory provisions apply to salient arrays and whatever closely resembles or is relevantly similar to their members. Close resemblance or relevant similarity is not determined by coordinated activity or convention but by reason. The nature and existence of such relations would, on this conception, clearly be amenable to argument.

For a time, I was beguiled by such a view. Nevertheless, it should be resisted. The passage from social salience to sufficiently close (on the basis of reason) is ad hoc. It is a hybrid of the social and the rational. The view becomes even more ad hoc if deferential conventions—conventions to defer to experts on the application of rules—are admitted, as I shall argue in Section V that they should be. For, setting aside cases where the statute is interpreted or judicially applied in a new way, it then follows that sometimes how a statute applies is directly a matter of social convergence (socially salient arrays), sometimes it is indirectly a matter of social convergence (deferential conventions), and sometimes it is (largely) independent of social convergence.

The implications of the tentative remarks on social salience are that in a debate about whether the Canadian bylaw permits a drug store to close its doors to business at 10 P.M. but reopen sharply at 12:01 A.M., it is intelligible for both parties to make their claims. Under different circumstances, either of the claims may turn out to be true. Moreover, there seems to be a role for argument in resolving the issue. So there is nothing in the social salience view incompatible with the existence of disagreement about statutory application and the appropriateness of arguments in resolving that disagreement.

It is social salience, rather than knowledge of independent legislative intentions or genuine kinds, that enables speakers to converge on salient act-tokens in the appropriate contexts. This tidies up the possibility of epistemic guidance in the simplest cases. Epistemic guidance—in these cases—now requires that to the conditions outlined at the outset (the promulgation of a statute, a minimal semantic and syntactic competence of the addressees) only one thing be added: a social competence among addressees that enables its possessor to recognize members of what in fact are the socially salient arrays. (Note that though these act-tokens will strike an addressee as salient, he need not think of them as socially salient.) If epistemic guidance entails that the one guided knows that the statute picks out the act-tokens in question, then epistemic guidance is going to be possible among the socially competent.⁵³

They will have correct beliefs about the relevant act-tokens, and those beliefs will be justified provided that these people have (sufficient) reason to believe themselves to be well socialized. Many people will have this justification, as their background social interactions will have given them an abundance of evidence. If social salience depends at any point on natural tendencies of mind and perception, many people will have ample evidence that they perceive and think normally. Thus, under normal social conditions, epistemic guidance need not in practice require anything more than the promulgation of a statute. This is because in the simplest case, the possibility of epistemic guidance rests entirely on sociolinguistic competences always present in a society. Moreover, the rationality of promulgating statutes—provided there are always simplest cases—follows automatically. (This last point will be developed more fully in the next section.)

V

I have been suggesting that the plain application of a statutory directive is its application in the simplest case and is given by the socially salient array for that directive. If one wants to think of plain meaning narrowly as determined by a combination of semantics and syntactical rules, then plain application is not determined by plain meaning alone but by the broader social environment. One consequence of this view seems to be that a statute that says that “those who sell gold jewelry” must obtain a certain kind of license applies to socially salient examples of people selling gold jewelry. (However, this consequence does not obtain under all conditions, as subsequent discussion will reveal.) If the society in which this statute is promulgated is one in which the atomic structure of gold has not been discovered, the statute will apply to people selling what is taken to be gold jewelry, even if some of it is not really gold. This sort of result is the most striking difference between social salience theory and applied semantic realism. It represents a refusal to account for a statutory directive's reference the way one might account for the reference of an empirical assertion or a scientific claim. And there are benefits to refusing, as I shall argue, below.

First, though, an objection from judicial practice should be considered. Judges and other officials have well-established practices of applying statutes in ways that seem independent of social salience. To give but a few examples: officials uniformly read a self-defense exception into homicide statutes. American judges interpret statutes in the light of legislative statements made a part of the record but not of the statute itself.⁵⁴

Courts apply certain statutes—for example, statutes criminalizing possession of controlled substances—only when the substance in question is or appears to experts to be a member of the genuine kind named in the statute, regardless of the arrays salient to the target population.

Some of these practices may pose no challenge to social salience theory. On the plausible assumption that it is widely known that self-defense is accepted in a charge of homicide, the existence of that practice is part of the background social context determining socially salient arrays for the target population. Moreover, whenever the issue is what judges do, a counterexample must be one of statutory application and not something related. If the distinctions suggested at the beginning of this article—between questions of statutory application, legal requirements, and proper interpretation of statutes—are sound, then some of the judicial practices alluded to may fall into a category other than the category of (pure) statutory application.

However, it is all too convenient to claim that in all such practices, something other than pure application of statutes is going on—convenient but not plausible. One wants to protest that in at least some such practices, the court is not making the statute apply this way by this practice but, rather, is recognizing how the statute (antecedently) applies. What becomes of social salience theory then?

A plausible example is the deference to experts on whether prohibited substances are possessed or sold. The esoteric knowledge of biochemists and the like is not widely shared. Suppose that members of the target population see the name of a particular controlled substance as standing for one set of things and experts identify another set. To the target population it is salient that the statute applies in certain instances; the experts claim it does not. Surely, it will be maintained, the experts can be correct.

Applied semantic realism has no difficulty explaining why the experts are correct, provided they know about the natural world they are describing. The intentionalist handles this example by claiming that it was the intention of the legislature to defer to experts to determine the presence of the prohibited substance. But social salience theory begins by noting that rather than being a paradigm of application of a statute, this situation represents an irregular case, one problematic for the success of epistemic guidance. By hypothesis, socially salient arrays exist and are incompatible with the applications experts, and through deference to them, officials recognize.

Fortunately, this way of conceiving of the statutory example is artificial. Some philosophers of language, following Putnam, maintain that there are deferential conventions among English speakers for terms like “cocaine,” conventions to defer to experts on the referents of those terms.⁵⁵

It is doubtful that socially salient arrays containing instances of the possession of cocaine both independent of and contrary to the input of experts exist to any significant degree among the target population of possession statutes. (There may, however, be what Putnam calls “stereotypes.”) If anything is salient, it is that instances of cocaine (and therefore, cocaine possession) are to be determined by experts. This deferential convention is reinforced by widely known criminal justice practices, such as the way strict construction of criminal statutes is practiced in common-law countries by courts, prosecutors, and grand juries. There may, of course, be socially salient arrays arising after and because of the input of experts.⁵⁶ But it would be the determinations of experts, not the socially salient arrays, that decide the applications of the possession statute.

The point about deferential conventions may be thought to show that social salience theory must be rejected, but it does not. What it does show is that the social salience account needs to allow explicitly for the use of deferential conventions. It can do so in a way that is not ad hoc. The claim has been that application of statutory provisions is determined by socially salient arrays, and the suggestion is that these are often the result of coordinated cooperative activity among speakers to whom the arrays are socially salient. The coordinating activity yields shared groupings of act-tokens for directives. At its simplest, the result of such ongoing activity is that people in the target audience of a directive come to share the perception that it applies in certain cases and can use this perception to guide their behavior to comply with the directive. But the picture can be complicated. Sometimes, what the coordinating cooperative activity produces is a deferential convention. When there are deferential conventions, the possibility of epistemic guidance depends upon the promulgation of rules of application by the experts, some heuristic device useable by members of the target population, or case-by-case direction.

It may seem as if statutory application in familiar, mature legal systems uses many deferential conventions. Suppose that instead of being the exception, deferential conventions are the rule (assuming this claim is coherent). Then the model of salient arrays would be of merely peripheral interest. But the significance of salient arrays cannot be dismissed in that way. It is true that in statutory interpretation at the appellate level, one finds in America the practice of consulting legislative records other than the statutory text. It is but a short hop to the conclusion that this is done in order to determine legislative intent. However, as I remarked at the outset, appellate cases applying (or interpreting?) statutes seem to be but a small percentage of statutory applications. The vast majority of statutory application is not done by judges and never results in a legal case; and for these statutory applications, there is no practice of consulting legislative records (and the consultation of empirical experts is not the rule).

But perhaps such consultation is indirect. Many of the people who apply statutes in everyday life rely on glosses of statutory provisions rather than statutory provisions themselves. We speak here not only of minor officials such as police officers, motor vehicle inspectors, town clerks, and building inspectors, but of laypersons such as driving instructors, building contractors, shopkeepers, and college admissions officers. The reliance on glosses occurs in everyday applications of sales tax statutes by merchants, motor vehicle ordinances by motorists, leash laws by dog owners, and so on. It is tempting to conceive of this practice as deference to the writers of the glosses, who, it may be claimed, rely in turn on other deferential conventions (consulting, for instance, evidence of independent legislative intentions or empirical experts).

One problem with this line is that the glosses themselves can epistemically guide. This capacity must be accounted for, and nothing in the story told so far has eliminated the apparatus of socially salient arrays in accounting for it. More to the point, it seems clear that what it takes to be an accurate gloss entails having the same application-meaning as the statutory provision glossed. So even in a system in which there is always deference to accurate glosses of statutory provisions, the application-meaning of a gloss is identical to the application-meaning of the relevant statutory provision. If the latter is determined by socially salient arrays (and even on the theory being proposed, that is not always the case),⁵⁷

the former is determined by those same arrays. If, that is, the view that socially salient arrays determine statutory application in the simplest case is sound, the conventional deference to glosses, when proper (i.e., when the glosses are accurate), is transparent in the simplest case to socially salient arrays and does not supplant them.

The point is that incorporating deferential conventions into the social salience perspective opens the door neither to abandoning socially salient arrays nor to making social salience of peripheral importance.

Deferential conventions introduce a subtlety and richness into the theory of statutory application that at times mimics applied semantic realism. For deferential conventions can make scientific disputes—for example, about the biochemical structure of a particular substance possessed by a particular defendant—directly relevant to statutory application. So if the experts in question are credentialed scientists, scientific disagreement about the real nature of a particular substance will be relevant to the application of a statute concerning the substance. Under certain circumstances, the relevant disagreement may even extend to the views of mavericks who are not in some sense commonly recognized as experts and who challenge the conventional wisdom. For the dispute in question need not be among mainstream experts alone; where a deferential convention requires only a particular kind of credential, a credentialed person whose views are out of the mainstream will qualify.

Although these results will please the applied semantic realist, the reason for the results will not. The disagreement about the real nature of a substance and the accompanying experimental data are relevant only, on social salience theory, because of the configuration of a particular deferential convention. More precisely, all of the experimental data are relevant for the question of pure statutory application only because of that convention. If the issue is whether a particular defendant has violated a criminal law by possessing or selling a substance, the data of maverick scientists about the substance actually possessed may be relevant for other reasons (e.g., because the data conflict with claims that the substance had a certain nature, and there are policies that give the benefit of doubt or disagreement about empirical facts to the defendant.)⁵⁸

Consider, next, the consequences of the full theory for diachronic stability in application-meaning. Salient act-tokens of a directive are salient relative to the social background against which the directive is communicated. This permits a different account of “enlightened” application of an elderly statute from one produced by applied semantic realists. Brink has us consider a hypothetical 1945 statute imposing strict standards in the handling of “toxic substances.” He supposes that the original legislators had in mind certain specific applications. With the passage of time, however, “we” have reason to believe these beliefs are mistaken. There is now a better theory about what things are, in fact, toxic. So we should apply the statute in a new and enlightened way.⁵⁹

Social salience theory can agree that the statute now applies in a new way, not the way the 1945 legislators assumed. It applies in a new way if advances in chemical knowledge have become salient or if it is a convention that experts are to be deferred to about such matters. Unlike applied semantic realism, however, social salience theory need not say that in 1945, those who thought it applied to material we now know to be nontoxic or denied it applied to material we now know to be toxic were necessarily incorrect. This result will be startling or not, depending on whether it is plausible that the correct application of statutory provisions can evolve over time, as some claim is true of Constitutional provisions.

The example of the toxicity statute illustrates a difference between social salience theory and an imputed legislative intention view to the effect that the legislature intends whatever it is reasonable to suppose their words communicate, given the linguistic and social context at the time of their initial communication. The latter view, which identifies application-meaning with a kind of imputed legislative intention, will allow for a change of application-meaning under some conditions—for example, when the intention imputed is a goal and a change in social circumstances changes the way the goal can be achieved. But if the imputed intention of the toxicity statute is to protect public health from substances that are actually harmful, there will be no change in application-meaning from 1945 to the present.

The view proposed here is deeply social. The default position on statutory application is that a statutory provision applies to members of socially salient arrays and does not apply to members of peripheral arrays. The role of the real nature of things in the world, the intentions of legislators, and so on figure in pure statutory application, when and where they do, only insofar as they affect salience or are incorporated into deferential conventions. Statutes are capable of epistemic guidance because of the default position and because target populations generally have mastered the necessary social and linguistic competences. (Target populations will be aware of and participate in the deferential conventions, when they exist, as well; however, more must be added when deferential conventions exist to make epistemic guidance possible.)

Social salience theory accounts for clear intuitions in the statutory examples given in Section III. As has already been pointed out, the theory can offer a simpler account of epistemic guidance than intentionalism (and perhaps than applied semantic realism). A related point is perhaps more important: compared to its competitors, social salience theory does a better job of justifying the practice of statutory promulgation, since it supports a claim about the rationality of this practice that extends to more situations. Finally, social salience theory can support a story that gives maximum credence to the idea that statutes are useful things for societies.

The discussion of the last two points begins on previously examined ground. Applied semantic realism makes the possibility of knowing the practical import of statutes depend upon factors other than what people tend to share in a society, such as the ability correctly to group things with the same real natures or properties (in addition to the relevance of named concepts to practical import). Intentionalism also makes knowledge of the practical import depend upon additional factors, as the burden of Section I discloses. Social salience theory does not, at least in the simplest case, when there are no deferential conventions for statutes. Nearly everyone to a large degree participates in some social practices of salient array-identification. It would not be an exaggeration to claim that this participation is necessary for full linguistic competence.⁶⁰

So when deferential conventions are not a factor, according to social salience theory, statutory practical import depends entirely upon what people tend to share in a society. It follows that, setting deferential conventions aside, once statutes are accessible, their practical import is accessible. (All of this is meant to apply to situations in which statutory application has not been subsequently changed—e.g., by later judicial interpretation.)

Recall that the applied semantic realist may say that there is normally a certain amount of accessibility of practical import in people's familiarity with paradigms of named concepts. But the (virtual) guarantee of accessibility ends here for the applied semantic realist. In contrast, social salience theory does not limit (virtually) guaranteed accessibility of practical import to paradigms of named concepts. It may be socially salient that a particular rape prohibition, for instance, applies to a particular kind of aggressive sexual act without that act being a paradigm of rape.

In short, social salience theory provides a default picture of statutory directives as having particularly accessible practical import. It follows that promulgation of statutes by merely publishing what the statute says (in some sense of “says”) is rational, even for a public ignorant of the real natures of things or under the sway of some revisable paradigms. It also follows that promulgation of statutes is rational even when the public has no way of knowing independent legislative intentions or the probability that the legislature intends the applications that are salient. It is rational to make public only what is said when normal social background information and competences will ordinarily supply the rest of the information necessary for epistemic guidance. So social salience theory supports the rationality of the practice of promulgating statutes under more circumstances than do applied semantic realism and intentionalism.

Moreover, this accessible practical import, according to social salience theory, is one around which there is convergence and about which there is (typically) mutual knowledge. The conditions that establish accessibility, according to social salience theory, make for convergence. They may also make mutual knowledge of that convergence likely. Because people are participants in social practices of salient array-identification, they will tend to expect this convergence, though they may credit it to factors other than social convention (e.g., to some alleged metaphysical fact such as: ceasing business at a drug store for ten overnight hours is a closing of a drug store). This accessibility, convergence, and mutual knowledge exist because statutes take advantage of a kind of social intelligence shared by the socially competent in a society—their tendency to see certain act-tokens in certain situations as saliently indicated by a particular promulgated directive.⁶¹

When one adds this point to familiar reasons why guidance and coordination of behavior by statutes are advantageous to a society—their flexibility compared to custom, the need to coordinate behavior that would not otherwise be coordinated for a variety of reasons⁶²

—it follows that social salience theory gives maximum credence to the idea that the practice of statutory directives is a valuable one. That an account of statutory application provides grounds for a stronger justification for promulgating statutes and a high evaluation of their social usefulness under broader circumstances is no reason to believe the account is correct, of course. But it is a happy consequence of that account and may serve to make it attractive.

Out of an abundance of caution, let us revisit the charge of vitiating legislative authority. If authoritative directives in the simplest case are to be understood in accordance with socially salient arrays, legislative authority may appear to be vitiated. For the addressees seem to have disproportionate power to determine the salient arrays. Alexander's complaint is that the “interpreters”—here, the addressees, who jointly find some arrays salient—are the real practical authorities. However, this is appearance, not reality. What is socially salient is generally not salient merely to the addressees. It is a rare case when the legislators do not share with the target population at the time dispositions to find certain applications salient. It is rarer still that the addressees are in a position to determine what will be socially salient after the legislature has acted (and so, under conditions that cannot be taken into account by legislative drafters), unless a considerable time passes between the passage of the statute and the application by addressees.

The problem or its appearance arises most acutely with the elderly statute and its applications under psychosocial conditions alien to the original legislature. What is socially salient to the later generations, let us suppose, is now radically different from what was socially salient at the time of the passage of the statute. If, as social salience theory directs, we opt for the current socially salient arrays to answer the question, “To which acts does this statute now apply?” it may initially seem that the original legislature's authority is undermined. But the objection is built upon a questionable assumption, namely that the original legislature's due authority in these matters survives. And even if this assumption is granted, it does not follow that the legislature's moral authority survives with respect to the directive in a way that makes the application of the directive determined by original legislative intentions. Intuitions on these matters are murky at best. Where these intuitions exist, they are controversial and not compelling. Hence I doubt that this sort of example can be used to undermine social salience theory decisively.

VI

The two previous sections have offered an account of the application-meaning of statutes in terms of socially salient arrays or deferential conventions, both of which are often known to people who have the competence to discern the “plain meaning” of a statute or statutory provision. The social salience theory may thus seem reminiscent of the “plain meaning” school of statutory interpretation, but there are important differences. The first is that if interpretation is assigning a meaning, then when socially salient arrays are used, there is no interpretation going on at all. The second is that, unlike many plain-meaning theorists, I am not making a prescriptive claim about how appellate judges should proceed in adjudicating questions about statutes. Unlike the so-called “textualism” espoused by Justice Scalia, social salience theory makes no claims about the application of Constitutional provisions; still less does it fix the date of the relevant arrays as the date of the original legislation. Third, “plain meaning” is an elastic term that can stretch to cover both sense and reference or shrink to cover just one of them. Social salience theory is about the analogue, for statutes, of reference only and only for those statutes or parts of statutes that purport to direct actions.

The social salience account is reminiscent of and more closely aligned to a position held by Hart. Hart claims in The Concept of Law that there is a “core of certainty” to rules that accounts for rules having “clear central” or “plain” cases of application. This core of certainty results from a “widely diffused capacity to recognize instances of the general classifications” used by a rule.⁶³

There is in Hart's 1961 work no implication that the core of certainty is created entirely by an acontextual core meaning of concept words used in the rule. On the contrary, Bix takes Hart to hold that the context of the general classifying terms is a factor in determining the core of certainty of a formulated rule.⁶⁴

The central ideas of social salience theory are that people have a capacity to recognize instances governed by formulated rules, given the context of the words used, and that upon this capacity depends the application of statutes in the simplest cases. So why not adopt what Hart says and be done with it?

The main reason is that talk of recognizing instances of general classifications is misleading on two counts. It is misleading because the implicit reference to the context of the classificatory words is indeterminate. There is nothing to correct familiar assumptions that the statutory linguistic context (the entire statute, its preamble, and its title), the legal context (background legal norms), or the legislative context (legislative intention and purposes, the legislative record), individually or in combination, is not only a context that counts but the context that counts. Some of these assumptions, I have been arguing, ought to be rejected. They are both too broad (legislative intention and purposes and often background legal norms do not count) and too narrow (background social practices and norms do).

The talk of recognizing instances of general classifications is also misleading in another way. Despite having begun writing about the classifications “which the law makes,” in the two pages following the quoted passage, Hart launches into talk about “verbal descriptions” and “clear verbal terms.” This language is applicable to rule formulations (as in statutes), not rules. As I have been arguing, the ability to recognize clear instances to which the verbal terms used by a statute apply is not the ability that counts for recognizing applications of the statute. Consider a statute imagined by Fuller: “It shall be a misdemeanor, punishable by a fine of five dollars, to sleep in any railway station.”⁶⁵

Those who understand the language and read the statute have the capacity to identify instances of sleeping in railway stations, including (1) the seated sleeping of a passenger waiting on a bench in the wee hours for a delayed train; and (2) the reclined sleeping of a homeless man who has made himself comfortable for the night. The problem is not to explain the ability to recognize instances of sleeping in railway stations, but to explain the ability to recognize instances of sleeping in railway stations that are forbidden by the statute. One might envisage social backgrounds in which both instances are forbidden. However, if, as Fuller implies, only the second instance is forbidden, something other than the capacity to recognize instances of general classificatory terms used in a statute (in their linguistic context) selects from the larger set of putative referents. The discussion of socially salient arrays calls attention to the mechanism that chooses within this set and directs attention to the subset that counts (or in some cases, such as those discussed in note 41, chooses outside the set). Hart's language invites us to overlook the existence of this mechanism.

Finally, my view ought to be distinguished from views that hold that it is the legislature's imputed intention that determines statutory application. In Section V, I indicate the difference between social salience theory and an imputed legislative intention view to the effect that the legislature intends whatever it is reasonable to suppose its words communicated at the time of the initial communication. Consider now a variant: the theory that the legislative intention to be imputed is an intention to apply the statute in accordance with social salience (or deferential conventions) at the time of the application. This intention can be imputed, on this view, on the basis of the manner of promulgation (the fact that in some sense of “said,” all that is presented is what the legislature said) and the common knowledge that this directive may be so durable that it remains promulgated in eras to come (and thus long after one can learn about the initial social context or specific concerns and desires of the actual legislators and long after what is salient has shifted).⁶⁶

Obviously, if such a view allows the imputations to be defeated (e.g., on the basis of the kind of evidence that the intentionalist would use, including additional communications), it differs from social salience theory. But why not hold that such imputations cannot be defeated?

I think this would be a tactical, if not a philosophical, error. The immunization of imputed intention from defeat seems arbitrary and calls for argument. I cannot think of an argument that would not be an argument solely for the middle term of the account, socially salient arrays (and deferential conventions). Further, legislative-intention talk, however carefully prefaced with “imputed,” invites the philosophically unsophisticated or confused to slide into something approaching intentionalism. Hence it seems best to reject “imputed intention” language.