Do theories of linguistic vagueness have implications for legal theory? That is, do particular accounts of phenomena such as the sorites paradox and borderline cases have implications for debates about the effects of vagueness on interpretation and adjudication, the role of vagueness in legal texts, legal indeterminacy, or the nature and fundamental grounds of law?
After an extended overview of his own theory of vagueness, Schiffer (Reference Schiffer2001) concludes “no”: “having done a bit of homework, I have reached the conclusion that philosophical theories of vagueness, even if true, have nothing to offer jurisprudential concerns about vagueness.”Footnote 1 Yet most have agreed that what theory of vagueness is correct does have implications for philosophy of law.Footnote 2 Facts about vagueness as understood in particular theories may be treated as placing nontrivial constraints on legal theorizing; alternatively, facts about legal discourse, practice, and interpretation may be invoked as grounds for adjudicating among theories of vagueness. Here is Endicott:Footnote 3
Vague laws … pose problems for philosophy of law.
Philosophical approaches to the [sorites] paradox seem to have implications for legal theory: arguments that vague terms are incoherent, and that reasoning with them is impossible, would support arguments that vague laws are incoherent. Since vague laws are an important part of every legal system [Endicott 2001], the implications seem to be far-reaching.Footnote 4
I disagree.
I focus on four putative implications discussed in the literature concerning:
• the value of vagueness in the law (Section I)
• the possibility and value of legal indeterminacy (Section II)
• the possibility of the rule of law (Section III)
• “strong” discretion (Section IV)
I argue that the implications of theories of vagueness on these issues are less “far-reaching” than often thought. Contrary to certain assumptions and arguments in the literature:
• Epistemicism can capture the value of using vague language to delegate authority over hard cases to future adjudicators.
• One's theory about the linguistic (in)determinacy of statements about borderline cases underdetermines one's theory about the possibility of indeterminacy in the law.
• If linguistic vagueness threatens the rule of law, it does so regardless of what theory of vagueness, theory of legal content determination, or conception of the ideal of the rule of law is correct.
• One's theory about the linguistic (in)determinacy of statements about borderline cases underdetermines one's theory about strong discretion and whether exercises of discretion necessarily involve changing the law.
I conclude with some methodological remarks (Section V).
Before getting started, some brief background on vagueness may be useful. Informally put, linguistic vagueness is an apparent fuzziness in the proper application of an expression. Ostensively, vague expressions are expressions such as tall, bald, rich, etc. Characteristic vagueness phenomena include sorites-susceptibility, tolerance, and apparent borderline cases. Even when all the relevant facts are in, we may be hard-pressed to say whether a man who is 5′10″ is tall (for a man). Such “borderline cases,” and the intuition that small changes in height do not induce changes in whether one is tall (“tolerance”), can lead to the sorites paradox. An application to the law from Endicott is as follows (where x n is a tire with n fewer molecules of rubber in its tread than a new tire):Footnote 5
(1) Sorites paradox (application to the law)
(P1) A new tire is not bald.
(P2) If a tire is not bald, it does not become bald by losing one molecule of rubber from its tread.
(C) So a tire never goes bald.
(C′) So no one can ever break the rule against careless driving by driving with bald tires.
(2) (P1) x 0 is not bald.
(P2) For all n, if x n is not bald, then x n+1 is not bald.
(C) So, for all n, x n is not bald.
(C′) So, a rule forbidding driving with bald tires can never be violated.
The premises seem true, and the argument seems valid.Footnote 6 But the conclusion is false. The challenge for theories of vagueness is to explain where the argument goes wrong and yet why it seems so compelling.
I use labels such as “sorites-susceptible,” “borderline cases,” etc., as descriptive labels for the above sorts of linguistic phenomena. My usage does not presuppose a particular analysis, or that the paradox is irresolvable, that borderline cases generate truth-value gaps, or even that linguistic vagueness is fundamentally semantic.Footnote 7 It is important to distinguish the notion of linguistic vagueness from notions such as ambiguity,Footnote 8 context-sensitivity,Footnote 9 multidimensionality,Footnote 10 and (metaphysical, metasemantic) indeterminacy.Footnote 11 It is a substantive question how linguistic vagueness may lead to indeterminacy in semantic or asserted content, and how vagueness in legal texts may lead to indeterminacy in the law.Footnote 12 The distinction between linguistic vagueness and indeterminacy in legal content will be important throughout the discussion.
I. CASE I: THE VALUE OF VAGUENESS IN THE LAW
I begin with an argument from Scott Soames that the debate between epistemicist and partial-predicate contextualist theories of vagueness has implications regarding the value of vague language in legal texts.Footnote 13 The upshot:
I conclude that whereas the genuine value of vagueness in the law is naturally explainable on the theory that treats vagueness as a matter of partial definition and context sensitivity, it cannot adequately be accommodated by the epistemic theory of vagueness.Footnote 14
Epistemicism claims that intuitively vague expressions P have precise but unknowable meanings and extensions.Footnote 15 Vagueness phenomena are symptoms of our ignorance of expressions’ precise extensions—e.g., ignorance of the precise standard for counting as tall, the precise cutoff between individuals that are bald versus not bald, etc. Inductive premises such as (P2) in (1)–(2) are false, and their negations—that there is a “sharp boundary” between the P and not P individuals—are true. Borderline cases b are items such that it is impossible to know whether b is P is true or false.
By contrast, theories such as partial-predicate contextualism deny that the linguistic rules and nonlinguistic facts determine precise extensions for vague expressions in context.Footnote 16 Vague predicates P denote partial functions partitioning a domain into a positive extension pos c(P), a negative extension neg c(P), and an extension gap gap c(P) (the “borderline cases”). However, in discourse speakers can “precisify” P by narrowing the range of items b ∈ gap c(P) for which P is conventionally undefined. Speakers can truthfully “go either way”—stipulate b ∈ pos c(P) or stipulate b ∈ neg c(P)—depending on their conversational purposes.
Critical cases for distinguishing the theories are certain core borderline cases—in particular, Soames claims, borderline cases that are more like things known to be in the predicate's extension. In short, Soames argues thus: epistemicism cannot capture the value of justifying different verdicts for such borderline cases given different legislative rationales; partial-predicate contextualism can. QED.
Soames's argument is as follows. Consider two variants of Hart's case of an ordinance No vehicles in the park.Footnote 17 In Scenario 1, the rationale of the law is to reduce pollution in the park, which has recently been disturbed by cars, motorcycles, etc. In Scenario 2, the rationale of the law is to reduce overcrowding, which has recently led to accidents on the roads and paths. Suppose a judge must render a verdict regarding a core borderline case of vehicle—an item b such that the truth-value of b is a vehicle is unknowable, per epistemicism, or conventionally undefined, per partial-predicate contextualism—say, a skateboard. Assume that judicial adjudication is “guided by a principle of fidelity to the law that assigns priority to maintaining existing legal content when possible, while mandating decisions that further legislative rationale when cases cannot rationally be decided on the basis of existing content alone”; and assume that “legal content” is the assertive or stipulative content of authoritative legal texts.Footnote 18
Partial-predicate contextualism warrants different verdicts in the two scenarios given the different legislative rationales. Vehicle as used by the lawmakers is undefined for borderline cases b such as skateboards. So the existing content of the law implies neither Skateboards are vehicles nor Skateboards are not vehicles. Since the case “cannot rationally be decided on the basis of existing content alone,” the decision must be made on the basis of the legislative rationale—i.e., whether forbidding skateboards would further reduce pollution or overcrowding on the paths, respectively. So in Scenario 1 the judge can truthfully assert Skateboards are not vehicles; she can precisify by stipulating that skateboards are in neg c(vehicle), making it the case that henceforth the content of the law does not forbid skateboards in the park. In Scenario 2 the judge can truthfully assert Skateboards are vehicles; she can precisify by stipulating that skateboards are in pos c(vehicle), making it the case that henceforth the content of the law implies that skateboards are forbidden in the park.
Being able to justify different verdicts in this way highlights a valuable function of vague language in legal texts. Lawmakers may be divided on how to classify certain borderline cases. Future adjudicators may have additional evidence about how certain classifications may promote or hinder the law's general rationale. By formulating the law using vague language the lawmakers can delegate authority over difficult and contentious cases to those in a better epistemic position to decide in light of the full facts of particular cases: “Hence the value of vagueness in the law.”Footnote 19
Soames argues that epistemicism cannot similarly justify different decisions in the two scenarios and hence cannot capture the valuable delegating function of vague language in legal texts:
[S]ince the content of the statute already determines the legal status of every borderline case, the first duty of the downstream authorities is to assign the borderline cases that come before them the legal status those items most probably already have.
[Since skateboards are] more probably vehicles than not … the epistemicist cannot justify arriving at different verdicts in the two scenarios.
In this way, the epistemicist's view of what vagueness really is prevents him from recognizing much of the value that vagueness in the law really has.Footnote 20
According to epistemicism, there is a determinate but unknowable fact about whether skateboards are in the positive extension or negative extension of vehicle, and thus whether Skateboards are vehicles is true or false in the context. So there is a determinate but unknowable fact about whether the content of the law implies that skateboards are forbidden in the park. Skateboards are more like things known to be in the positive extension of vehicle (e.g., a car) than things known to be in the negative extension of vehicle (e.g., the number 7). So it is more likely that the law implies that skateboards are vehicles. So fidelity to the law requires the same verdict regardless of the legislative rationale. So lawmakers cannot reasonably use vague language with the expectation that future adjudicators will decide in ways that best promote the legislative rationale.
Key assumptions of Soames's argument are that speakers’ rationales in using an expression cannot provide evidence about the expression's extension, and that how “similar” objects count as being in a context is independent of factors such as speakers’ rationales. These assumptions are questionable.
Epistemicists treat an expression's meaning and extension as depending (at least in part) on facts about use. Such usage facts include facts about the contexts in which the expression is (is not) used, speakers’ verbal dispositions to use (not use) the expression, speakers’ linguistic and extralinguistic goals when using or disposed to use the expression, and so on. One might wonder why the fact that vehicle was used in a context c with an intention to reduce overcrowding/accidents could not be among the usage facts that might, for all we know, help determine the predicate's extension, i.e., so that skateboards are in pos c(vehicle). Indeed a crucial component of Williamson's account of why vague expressions’ precise extensions are unknowable is that they are—to use a phrase from HawthorneFootnote 21—semantically plastic: Slight changes in use can lead to slight changes in vague predicates’ meaning and extension (cutoff, standard); in nearby possibilities where the usage facts differ only slightly, the meanings and extensions differ as well.Footnote 22
Given this epistemicist metasemantics and epistemology, the lawmakers’ rationale in using vehicle may, for all we know, be among the facts determining the predicate's extension and hence the content of the ordinance. So evidence about the lawmakers’ rationale may, for all we know, constitute evidence about whether skateboards are in pos c(vehicle), and hence whether skateboards are forbidden in the park. So evidence for different rationales may justify different verdicts. Far from being incompatible with this idea, epistemicism seems to positively predict a role for considering legislative rationales in certain core borderline cases—even given Soames's assumed norm of adjudication that “assigns priority to maintaining existing legal content.”Footnote 23 One can capture the value of “employ[ing] vague language as a way of delegating authority over difficult cases” without diagnosing the process in terms of “incremental, case-by-case precisification” of legal content.Footnote 24
Of course one might object to epistemicism's metasemantics and commitments about our irresolvable semantic ignorance. What is important here is simply that Soames's argument fails to provide an independent argument against epistemicism. Whether vague predicates are conventionally undefined for certain borderline cases may be less relevant to issues about the value of vague language in the law than it initially seemed.
II. CASE II: INDETERMINACY IN THE LAW
It might seem obvious that what theory of vagueness is correct has implications regarding the possibility of legal indeterminacy. Whether the linguistic facts about the truth-values of sentences about borderline cases are determinate—e.g., whether Skateboards are vehicles is determinately true or false—would seem directly relevant to whether the legal facts about such cases are correspondingly determinate—e.g., whether skateboards are forbidden in the park. Hence Endicott writes, “[epistemicist theories] imply that there is always a right answer to the application of a law stated in vague language;”Footnote 25 “If it [i.e., epistemicism] succeeds, then the indeterminacy claim [(3)] is false.”Footnote 26
(3) The Indeterminacy Claim (Endicott)
a. “[V]agueness leads to indeterminacy in the law.”Footnote 27
b. “[T]he law is indeterminate when there is no single right answer to a question of law, or to a question of the application of the law to the facts of a case.”Footnote 28
Since Endicott wishes to maintain that the law may be indeterminate on certain borderline cases, he rejects epistemicism:
(2) A legal theory should accept the indeterminacy claim.
… [C]laim (2) only holds up if vagueness leads to linguistic indeterminacy … Claim (2), therefore, relies on the rejection of the epistemic solution to the sorites paradox [i.e., epistemicism].Footnote 29
Epistemicism of course cannot grant (what Endicott calls) the “traditional formulation” of the indeterminacy claim, that “a vague statement is ‘neither true nor false’ in a borderline case.”Footnote 30 Yet epistemicism can still grant interesting senses in which vagueness may lead to legal indeterminacy—indeed that in decisions about certain borderline cases there is “no single right answer to a question of law” or to “whether the [vague] rule applies.”Footnote 31
What is at issue is how linguistic vagueness can lead to indeterminacy of content—specifically, how vague language in legal texts can lead to indeterminacy in the content of the law. To fix ideas suppose we understand “languages” in the manner of Lewis as formal objects that assign precise semantic values,Footnote 32 and let us individuate “words” (predicates, expressions) as lexical items (phrases) in languages thus understood. The epistemicist's metasemantics and epistemology of linguistic vagueness is understood as the claim that there is a determinate and unknowable fact about what (formally precise) language L is being spoken. Questions about borderline cases b are understood as questions about what lexical item is expressed by the relevant string of symbols/sounds—formally, whether the usage facts determine that the string ‘P’ is a predicate P 1 of a language L1 such that b ∈ ⟦P 1⟧L1, or determine that ‘P’ is a predicate P 2 of a language L2 such that b ∉ ⟦P 2⟧L2. (I will continue to use italics for expressions and reserve single quotation marks for strings. ⟦ ⟧L is the interpretation function, which assigns precise semantic values to expressions of L.)
One way in which the epistemicist might accommodate legal indeterminacy is by adopting the following sort of toy substantive legal norm and metaphysics/metasemantics of legal content determination:
(4) Indeterminacy-friendly epistemicist philosophy of law
In decisions about core borderline cases b of ‘P’:
a. Metasemantics/epistemology:
There is a determinate and unknowable fact about whether
(i) ‘P’ expresses a predicate P 1 of a language L1 such that b ∈ ⟦P 1⟧L1,
(ii) or ‘P’ expresses a predicate P 2 of a language L2 such that b ∉ ⟦P 2⟧L2.
b. Substantive legal principle:
(i) It is not impermissible for the judicial authority to proceed as if ‘P’ is a predicate P 1 of a language such as L1,
(ii) and it is not impermissible to proceed as if ‘P’ is a predicate P 2 of a language such as L2.
c. Metaphysics of content determination:
(i) Deciding ‘b is P’ makes it the case that L1 is now being spoken, and that the string ‘b is P’ now expresses a true statement about the content of the law. (In the post-decision context c, ‘b is P’ expresses b is P 1 and ⟦b is P 1⟧Lc is true.)
(ii) Deciding ‘b is not P’ makes it the case that L2 is now being spoken, and that the string ‘b is not P’ now expresses a true statement about the content of the law. (In the post-decision context c, ‘b is not P’ expresses b is not P 2 and ⟦b is not P 2⟧Lc is true.)
First, such an account captures the informal idea that there is “no single right answer” about how to decide on the basis of the existing legal materials. In the “No vehicles in the park” scenarios from Section I, one might decide ‘Skateboards are vehicles’ (given Rationale 2), making it the case that a language such as L1 is being spoken, that the string ‘vehicle’ expresses vehicle 1, and that skateboards are forbidden in the park. Or one might decide ‘Skateboards are not vehicles’ (given Rationale 1), making it the case that a language such as L2 is being spoken, that the string ‘vehicle’ expresses vehicle 2, and that skateboards are permitted. Second, the account captures ideas about the underdetermination of legal content. Since neither way of proceeding is legally excluded, extralegal considerations might provide the basis for a judge's decision. It is not the case that the content of the law post-decision is legally determined by the content of the law pre-decision. Indeed it is possible for a decision ‘b is not P’ to express a true claim about the content of the law in the post-decision context even if the string ‘b is P’ had expressed a truth in the (lawmakers’) pre-decision context. These substantive points about legal theory and practice are compatible with the epistemicist claim that there is a determinate linguistic fact about the lawmakers’ use of the relevant string ‘P’.
Two caveats: first, the substantive philosophy of law in (4b)–(4c) is not “anything goes”: it is not the case that anything a judge says is to be interpreted as true. The principles in (4b)–(4c) apply only to decisions about genuine borderline cases b—cases diagnosed alternatively as cases where it is determinate and unknowable whether b is in the (anti)extension of the predicate expressed by ‘P’, per epistemicism, or where b is in the set of items for which the predicate's semantic value is undefined, per partial-predicate theories. Of course it may be contentious whether a given item is a genuine borderline case. It is a substantive question of proper legal practice whether lawyers or judges ought to explicitly acknowledge an item's potential borderline status when defending clients or justifying decisions. Such issues are independent of whether epistemicism is correct, or whether the semantic/asserted content of the original legal text is determinate and unknowable versus undefined.
Second, the metaphysics in (4c) treats the adjudicator's linguistic act of uttering, say, ‘b is P’ as sufficient to make it the case that a given language (here L1) is being spoken and that the utterance expresses a truth about the content of the law. The condition could be strengthened depending on one's substantive views about hard cases—e.g., to require also that the judge's decision be precedent-setting in a certain way. Such issues regarding the correct metaphysics are independent of whether epistemicism about linguistic vagueness is correct. For instance, in a partial-predicate contextualist theory (Section I), where P is the partially defined predicate determinately expressed by the string ‘P’, the above issue would be reframed in terms of whether the judge's assertion of b is P suffices to make it the case that b is in pos c(P), or whether b is still in gap c(P) unless certain further conditions obtain (e.g., that the decision is relevantly precedent-setting). For expository purposes I will stick with the simpler formulation in (4c).
III. CASE III: THE RULE OF LAW
Vagueness in the law raises a prima facie threat to the ideal of the rule of law. In the ideal:
[L]aws must be open, clear, coherent, prospective, and stable, legislation and executive action must be governed by laws with those characteristics, and there should be courts that impose the rule of law. The organizing principle of these requirements is, as Joseph Raz puts it, that ‘the law must be capable of guiding the behaviour of its subjects’.Footnote 33
A natural idea is that the law should not be arbitrary. Yet it is hard to see how a judge forced to make a decision about a borderline case or a cutoff in a sorites series could avoid being arbitrary. After all, the worry goes, as far as conventional meaning and one's evidence are concerned, a speaker may “go either way.”
Endicott takes seriously the challenge from vagueness for the rule of law.Footnote 34 Summarizing:
Philosophers of law … have debated the nature of borderline cases, and its implications for … the possibility of the rule of law. If the application of vague laws is indeterminate in some cases, then in those cases a judge (or other official) responsible for applying the law cannot do so (and in fact, no one can use the law to guide their conduct).Footnote 35
Vagueness may seem to pose a much worse puzzle [for the ideal of the rule of law] … [I]f the law is vague, judges are sometimes called on to resolve disputes for which the law provides no resolution. With respect to such disputes, the law does not constrain the will of the judge … So, when the law is expressed in vague language, the result is some degree of arbitrary government.Footnote 36
The challenge proceeds in two steps: (i) linguistic vagueness leads to legal indeterminacy; (ii) legal indeterminacy (prima facie) threatens the rule of law. Hence, Endicott notes, two general strategies of reply are (i) to accept legal indeterminacy and deny that it threatens the rule of law, such as by revising one's account of the ideal (Endicott's strategy); or (ii) to “reject [legal] indeterminacy. This could be done by denying that the application of vague language is indeterminate in a borderline case. The epistemic theory of vagueness makes that denial.”Footnote 37
Pace Endicott, if linguistic vagueness (borderline cases, sorites-susceptibility) threatens the possibility of the rule of law, it does so regardless of what theory of vagueness, theory of legal content determination, or conception of the ideal of the rule of law is correct. Linguistic vagueness can lead to apparent hard cases in which—depending on which overall theory is correct—the law is determinate but unknowable or genuinely indeterminate. In such cases it may appear to the public as if, no matter how the judge decides, the decision will be arbitrary. This appearance has the potential to threaten trust in the judicial system and undermine subjects’ willingness to regulate their actions in accordance with official pronouncements. The system may lose the confidence and deference of those bound by it; it may no longer foster the “dispositions and attitudes that lead to default obedience to law,” and hence no longer serve its “essential function” in “creat[ing] opportunity paths for constructive individual and shared action over time.”Footnote 38 In practice judicial authorities must be advised, implicitly or explicitly, about how to proceed and what sorts of considerations (e.g., moral) may and may not be cited as grounds for one's decision. Either the resulting judicial practice will—actually or expectably, reasonably or unreasonably—undermine the rule of law, or it will not. Whether, as a matter of metaphysical fact, there is genuine legal indeterminacy is irrelevant. Knowing that the law is determinate but irresolvably unknowable—alternatively, that the law is indeterminate but the ideal only involves not “depart[ing] from the reason of the law”Footnote 39—is of little comfort if one is worried about whether apparent borderline cases will undermine the legal system's ability to “guide the behaviour of its subjects.”Footnote 40
IV. CASE IV: STRONG DISCRETION
The final case I wish to consider concerns vagueness and Hartian discretion—“strong” discretion in the sense of Dworkin's “The Model of Rules I”:Footnote 41
[I]f a case is not clearly covered by an existing legal rule, … [e.g.] because the rule contains vague or ambiguous terms, the deciding judge cannot apply the law but must exercise his or her discretion to resolve the case. Call this the Discretion Thesis.Footnote 42
[In borderline cases the judge] will need to exercise strong discretion… If his decision is used as a precedent, he will make new law.Footnote 43
[I]n hard cases …, owing to the indeterminacy of legal rules …, judges are left with the discretion to make new law.Footnote 44
That is: (i) Vagueness and legal indeterminacy: There are hard cases for which the content of the law fails to imply a determinate verdict, e.g., due to vagueness. (ii) Discretion: Such cases can only be decided by applying extralegal standards, i.e., by exercising Hartian strong discretion (hereafter simply “discretion”). (iii) Legal change: Given the prior legal indeterminacy, such discretionary decisions change the law. Considering these issues will provide a useful way of bringing together points from the previous sections on epistemicist and non-epistemicist theories of vagueness, linguistic and legal content, and indeterminacy.
It is common to think that one's theory of vagueness has implications for one's commitments about discretion and whether discretionary decisions involve changing the content of the law.Footnote 45 The above theses about discretion are prima facie contextualist-friendly and epistemicist-unfriendly; here is Soames:
Since, on [partial-predicate contextualism], there simply is no fact of the matter about whether these borderline cases … are, or are not, vehicles, the court's inquiry must be directed toward other matters … Once such a precedent-setting judicial decision has been reached, the content of the law will change.
For epistemicism, … the effect of the decision on the content of the law [is] different… If the decision is precedent setting, the content of the law … will not change.Footnote 46
The thought is that insofar as epistemicism implies that the linguistic content of the legal texts and thus the content of the law are determinate, legal decisions in hard cases cannot be grounded in extralegal considerations and do not change the law; conversely, insofar as contextualism implies that the linguistic and hence legal content may be indeterminate, certain decisions may need to apply extralegal considerations and make new law. (One's take on the upshot of these implications would depend on one's commitments about vagueness and discretion (“one person's modus ponens…”).)
First, we have seen that epistemicism about linguistic vagueness is compatible with legal indeterminacy (Section II). The hypothetical epistemicist philosophy of law in (4) accepts that the linguistic content of a body of legal texts is determinate while allowing that the content of the law may be indeterminate. Accepting epistemicism is thus compatible with accepting that judges may exercise discretion in borderline cases, in the sense of making a legal decision that is not determined by the previous content of the law. Yet it is not immediately clear that what changes in a discretionary decision is necessarily the content of the law.
Recall the “No vehicles in the park” case from Section I. Suppose that the judge decides ‘Skateboards are vehicles’. According to the epistemicist legal theory from Section II, the judge's usage effects (inter alia) a metalinguistic change in which lexical item the string ‘vehicle’ expresses: in the post-decision context c, ‘vehicle’ expresses vehicle 1 and ⟦Skateboards are vehicles 1⟧Lc is true; the content of (e.g.) ‘Skateboards were not vehicles but now skateboards are vehicles’ is false.
(5) false: Skateboards were not vehicles 1 but now skateboards are vehicles 1.
Such examples may be understood analogously to examples such as (6) involving (actual or hypothetical) metalinguistic change.
(6) false: If we called tails ‘legs’, then horses would have one leg.
How many legs horses have does not depend on our linguistic practices. The counterfactual is interpreted by evaluating the content of the consequent as determined by the expressions’ actual meanings (i.e., the proposition that horses have one leg) at the relevant possibilities described by the antecedent (i.e., possibilities in which we use ‘leg’ the way we actually use ‘tail’). What is true is (7). Analogously, on the epistemicist legal theory under consideration, what is true in the judge's context of use is not (5) but (8).
(7) true: If we called tails ‘legs’, then the string ‘horses have one leg’ would express a true sentence in the language we would be speaking.
(8) true: The string ‘skateboards are vehicles’ expresses a true sentence, though it might not have expressed a true sentence in the pre-decision context/language.
It is instructive to contrast examples such as (5)–(8) with performative utterances. After sincerely uttering ‘I promise to ϕ’ one can truthfully say:
(9) true: Now I am obligated to ϕ, but I was not before uttering ‘I promise to ϕ’.
One's sincere utterance with the performative verb is what creates one's obligation. By contrast, the sentence expressed by ‘skateboards were not vehicles’ is false in the judge's context even if the string ‘skateboards are not vehicles’ may have expressed a true sentence in the pre-decision context. The judge's decision makes it the case that producing the symbols/sounds ‘skateboards are vehicles’ constitutes a use of a true sentence of the language being spoken. Exercises of strong discretion need not be understood as effecting changes in legal content.
Much of the discussion thus far has focused on showing that accepting an epistemicist theory of vagueness is, contrary to arguments in the literature, compatible with a range of views on the import of vagueness for legal theory and practice. In closing, it is worth observing that the opposite point holds as well: accepting a non-epistemicist theory of vagueness need not require commitments to (e.g.) legal indeterminacy or strong discretion.
To fix ideas consider a contextualist theory of vagueness such as developed in Silk's Discourse Contextualism (“vagueness as contextual indecision”).Footnote 47 As part of a general semantic/pragmatic framework of Discourse Contextualism, the theory accepts the following commitments regarding conventional meaning and use:
(10) Contextualist philosophy of language
a. Semantics: Compositional semantic derivations take as given a precise representation of context g c that provides semantic values for context-sensitive expressions, e.g., precise standards/cutoffs.
b. Pragmatics: Uses of ϕ presuppose that the concrete discourse context determines a representation (or family of representations) of context g c such that ⟦ϕ⟧gc is true.
c. Metasemantics: Different concrete discourse contexts may determine different representations of context, hence different standards/cutoffs.
d. Vagueness: Typical concrete contexts are compatible with a range of representations of context, hence a range of standards/cutoffs.
The semantics is “contextualist” in the sense that intuitively vague expressions are interpreted with respect to the same sort of contextual parameter as paradigm context-sensitive expressions (here a contextually supplied assignment function g cFootnote 48); as far as the conventions of the language go, different concrete contexts may supply different standards/cutoffs. Although derivations of conventional meanings utilize a particular formal representation providing semantic values for context-sensitive expressions, concrete discourses may fail to determine a unique representation of context. There may not be a unique formal representation that counts as representing a given concrete discourse. With uses of intuitively vague expressions there are typically a range of live standards/cutoffs compatible with the speakers’ commitments. Vagueness phenomena are diagnosed in terms of such “contextual indecision” about what standards/cutoffs to accept. For instance, a borderline case b of tall would be understood (very roughly) as a case where b is tall is accepted given some live standard/cutoff for counting as tall and rejected given some other live standard/cutoff. Different ways of resolving our indecision may give different verdicts about whether the predicate applies.Footnote 49
One way of extending this general framework to legal contexts and the law is as follows. Contrast the hypothetical epistemicist philosophy of law in (4) with the contextualism-based account in (11).
(11) Indeterminacy-/Discretion-unfriendly contextualist philosophy of law
a. Semantics and theory of vagueness:
Discourse Contextualism + “vagueness as contextual indecision” (see (10))
b. Legal metasemantics and content determination:
Concrete legal contexts determine unique representations of context, hence unique asserted contents for uses of intuitively vague expressions.
c. Substantive legal norms:
(i) In practice, judges are to pragmatically presuppose that their concrete context determines the same representation of context as the original concrete context of use.
(ii) In decisions about borderline cases, judges may not invoke independent moral considerations as a basis for decision.
Concerns about arbitrariness typically lead speakers not to presuppose precise standards in discourse. Yet sometimes our purposes can override such concerns. Avoiding incoherence or gross error, as in a “forced march” sorites (dynamically presented sorites series),Footnote 50 can be one such purpose. Purposes of legal practice, even maintaining the “rule of law” (Section III), might be another.Footnote 51 Accepting a substantive philosophy of law such as (11)—e.g., that purposes of legal discourse and practice call for presupposing that the concrete situation determines a stable representation of context that supplies standards/cutoffs for intuitively vague expressions—is compatible with accepting a contextualist theory of linguistic vagueness such as (10). One can reject strong discretion and accept an assumption of determinacy in legal contexts without thinking that such commitments are required for semantic competence or encoded in the conventions of the language.
One might wonder whether legal contexts are in fact distinguished from typical discourse contexts in such a way that warrants the general assumptions of determinacy in (11).Footnote 52 What is important for our purposes is simply that, as epistemicism is compatible with accepting legal indeterminacy and a role for invoking extralegal considerations in decisions about borderline cases, non-epistemicist theories are compatible with denying that vagueness leads to legal indeterminacy and that decisions on hard cases require exercising strong discretion. One's theory about the linguistic (in)determinacy of statements about borderline cases underdetermines one's account of discretion.
V. CONCLUSION
It is common to think that what philosophical theory of linguistic vagueness is correct has implications for debates in philosophy of law. As Endicott writes:
Philosophical approaches to the [sorites] paradox seem to have implications for legal theory… Since vague laws are an important part of every legal system …, the implications seem to be far-reaching.Footnote 53
This paper has argued that the implications of theories of vagueness for legal theory may be less “far-reaching” than often thought. I focused on four arguments in the literature concerning putative implications between theories of vagueness and (i) the value of vague language in legal texts, (ii) the possibility of legal indeterminacy, (iii) the possibility of the rule of law, and (iv) strong discretion. I argued that, contrary to certain arguments in the literature:
(i) Epistemicism can
– acknowledge a legal role for considering legislative rationales in decisions about certain core borderline cases, and
– capture the value of “employ[ing] vague language as a way of delegating authority over difficult cases” to future adjudicators in a better epistemic position to further the law's rationale.Footnote 54
(ii) Accepting linguistic determinacy about the meanings/extensions of vague sentences is compatible with accepting the possibility of legal indeterminacy. Accepting an epistemicist theory of vagueness is compatible with accepting a legal theory—substantive legal principles and a metaphysics of (legal) content determination—according to which, in decisions about certain core borderline cases,
– there may be “no single right answer” to the legal question of how to decide, and
– the content of the law pre-decision legally underdetermines the content of the law post-decision.
(iii) If linguistic vagueness threatens the rule of law, it does so regardless of whether the relevant linguistic or legal content is indeterminate versus determinate and unknowable.
(iv) One's theory about the linguistic (in)determinacy of statements about borderline cases underdetermines one's theory about strong discretion. For instance:
– Accepting epistemicism is compatible with accepting legal indeterminacy and a role for invoking extralegal considerations in decisions about certain borderline cases.
– Conversely, rejecting epistemicism in favor of (say) contextualism is compatible with denying, as substantive extrasemantic matters of legal theory and practice, theses of legal indeterminacy and discretion.
I want to be clear about what my argument is not. I am not denying that there is interesting practical and theoretical work to be done on issues of vagueness and the law.Footnote 55 Vagueness in legal texts raises difficult questions about (e.g.) how judges ought to proceed in decisions about apparent borderline cases; how to maintain the rule of law and public confidence and trust in the legal system; whether independent moral considerations may provide a basis for decision; and whether such moral considerations would count as part of the law and fundamentally determining legal content. One's linguistic diagnosis of vagueness phenomena—borderline cases, the sorites paradox, tolerance—will not help one make progress on such questions.
Much of our discussion has been critical, raising worries for arguments claiming implications of theories of vagueness for legal theory. But there is a constructive lesson. By delineating issues concerning the meaning, content, and function of vague language in the law, “the result can only be healthy for all … disciplines.”Footnote 56 Distinguishing questions about substantive legal norms, legal interpretation, and the metaphysics/metasemantics of (legal) content determination from the semantics proper can free up legal inquiry. This can motivate clearer answers and a more refined understanding of the space of overall theories of vagueness, interpretation, and law.Footnote 57