1. Historical introduction
Systematic efforts to use philosophical insights about language to solve problems in philosophy of law are relatively recent. Jeremy Bentham was perhaps the first to make a deliberate attempt at it. He developed a radically empiricist theory of the meaning of words, which supported his utilitarianism and his legal theory. What is the meaning of lawyer.
Bentham wanted to abandon what he considered to be a nonsensical mythology of natural rights and duties—that is, moral rights and duties that people have regardless of whether anyone is prepared to enforce them. He looked for ‘sensible’ phenomena by which to explain the nature of law. Linguistic acts struck him as respectably empirical phenomena, and he made them an essential element of his theory of law. He based his ‘ legal positivism ’ on his claims about the meaning and use of words.
Language had not been especially important to the natural law theorists whose views Bentham despised. They accounted for a law as a certain sort of reason. From that perspective, philosophy of language has no special role in explaining the nature of law. Philosophy of language cannot explain the nature of reasons; it has the ancillary role of explaining the possibility of communicating reasons and the possibility of creating reasons by the use of language (so that natural law theorists have the same reason as others to seek a good account of the legal effects of the use of language, and of the role of courts in resolving disputes over it). Bentham, by contrast, tried to use the ‘sensible’ phenomenon of a perceptible, intelligible linguistic act for his purpose of expounding the nature of law by reference to empirical phenomena.
Bentham seems to have thought of the meaning of a word in causal terms, as its capacity to act on a subject by raising an image of perceptible substances or emotions for which, he said, the word was a name. ‘By these general terms or names, things and persons, acts, and so forth are brought to view…’ (Bentham 1782, 82; see also Bentham 1776, 28, 108n). Words that do not bring to view such perceptible things have no meaning, on his theory, except insofar as they can be expounded by ‘paraphrasis’—Bentham’s method of translating whole sentences in which those words are used into sentences that do raise images of perceptible things.
To many legal theorists that approach appeared, as H.L.A. Hart put it, ‘as a revelation, bringing down to earth an elusive notion and restating it in the same clear, hard, empirical terms as are used in science’ (Hart 2012, 84). The theory supported not only Bentham’s empiricism, but also his utilitarianism, because it privileged what he viewed as the ultimate sensible (and therefore intelligible) ‘affections’: the pain and pleasure that utilitarianism treats as the basis for a theory of value and of morality.
In his legal theory, this view of language became the basis of an innovative account of law as the expression of the will of a sovereign in a political community. Bentham stated it as follows:
A law may be defined as an assemblage of signs declarative of a volition conceived or adopted by the sovereign in a state, concerning the conduct to be observed in a certain case by a certain person or class of persons, who in the case in question are or are supposed to be subject to his power… (Bentham 1782, 1)
He went on to explain that the signification of volition by such an assemblage of signs must be backed by ‘motives’ of pain or pleasure offered by the sovereign.
Two features of this theory tie the philosophy of law to the philosophy of language. One feature is methodological, and the other is substantive. First, Bentham proposes his theory as a definition of the term a law (see section 3.1 below, on ‘Definition as a methodology in philosophy of law’). Secondly, he defines a law as a particular kind of assemblage of signs (see section 2.1 below, on ‘Law and Signs’). In Bentham’s view, a law is an assemblage of signs, and legal philosophy is a form of philosophy of language. The legal theorist has a linguistic task of defining the terms (especially law, but others as well) of legal discourse.
That, in brief, is the apogee of the use of philosophy of language in philosophy of law. Bentham was ahead of his time. His theory of the meaning and use of words anticipated various trends in twentieth-century philosophy of language (including Frege’s and Wittgenstein’s ‘context principle’, some views of logical positivists, and the development of speech act theory). It was who, in the 1950s and 1960s, made a concerted effort to use twentieth-century developments in philosophy of language to ‘elucidate’ the nature of law. He did so with an enthusiasm for the work of Wittgenstein, and also of the Oxford philosophers J.L.Austin and Paul Grice. So Hart had some advantages over Bentham. Wittgenstein’s Philosophical Investigations had been directed against the view that a word’s meaning is the thing for which it stands as a name. That was one of the misconceptions that distorted Bentham’s theory of the meaning of words: he thought that a word like ‘right’ had to be a name for an entity, and that since no such entity could be sensed, the word was a name for a ‘fictitious entity’—(Bentham 1782, 251). And J.L.Austin took an attitude to ordinary discourse that was quite opposed to that of Bentham, who thought that philosophy must tear away the ‘veil of mystery’ that ordinary language throws over every object of study (Bentham 1782, 251). Wittgenstein’s attitude was more complex: he thought both that philosophers create philosophical problems by bewitching themselves with language, and also that a clear understanding of the use of language could provide healthy therapy for people suffering from philosophical problems. Taking advantage of the insights of Wittgenstein and J.L.Austin, Hart aimed to put philosophy of language to work in addressing problems of legal philosophy, without making what he regarded as Bentham’s extravagant mistakes.
In 1961, Hart’s book The Concept of Law raised issues that have occupied legal philosophers ever since. He borrowed J.L. Austin’s method of ‘using a sharpened awareness of words to sharpen our perception of the phenomena’ (Hart 2012, v, 14). Hart’s observations about the use of language in law were elements in an innovative approach to explaining the normativity of law—that is, the fact that law presents itself as conferring rights and powers and as imposing obligations and liabilities. Hart argued that we can understand that feature of law more clearly, if we understand where Bentham and his nineteenth-century disciple John Austin (not to be confused with J.L.Austin) went wrong in explaining the meaning and use of normative language. Hart’s new approach to the issue has been a starting point for discussions of the normativity of law since the 1960s [see section 3.2 below].
Ronald Dworkin argued that Hart’s focus on language had a toxic effect on his whole approach to legal philosophy. He wrote that Hart suffered from a ‘semantic sting’, because he wrongly thought ‘that lawyers all follow certain linguistic criteria for judging propositions of law’ (Dworkin 1986, 45; see section 2.4 below). That argument in the philosophy of language has set an agenda for much debate in philosophy of law (see, for example, the essays in Coleman, 2001).
Because of the allure of philosophy of language, and partly as a result of the influence of Hart, the use of philosophy of language in philosophy of law became the focus of much jurisprudential thought and debate in the 20th century. One notable example is the work of Jerzy Wróblewski, a Polish legal theorist who developed a ‘semantic conception of a legal norm’ (Wróblewski 1983), and offered a theory of legal language as a sub-type of natural language with multiple ‘fuzzy’ characteristics (Wróblewski 1985). In Italian analytical legal philosophy, the influential theorist Norberto Bobbio began in the 1950s ‘to consider law as a language and legal science as a meta-language’ (Pintore and Jori 1997, 3), in service of a theory of law based on a non-cognitivist moral theory. Two succeeding generations of Italian legal theorists have engaged in ‘dispute on semiotic ground, rather than directly by arguments of general legal philosophical positions’ (Pintore and Jori, 17).
That development was one instance of the widespread use of views about language in debunking conventional views about law in the 20th century. Perhaps this development resulted not so much from the influence of Bentham’s own writings, as from pursuit of the same search that he undertook, for ways of identifying the elements of law that suit empiricist philosophical sensibilities. The energetic 20th-century school of ‘Scandinavian legal realism’ viewed legal terms like ‘right’ as ‘lacking semantic reference’ and ‘denoting nothing’. So they considered statements asserting the existence of rights, duties, and other legal relations to be incapable of being true or false (Olivecrona 1971, 246, 255, 261). They variously explained the use of such statements as attempts to perform magical incantations, or as tools for taking advantage of the psychological conditioning that leads officials and citizens to act in one way or another when they hear such statements (see the outline of Scandinavian realism in Olivecrona 1971, 174–182, and see Ross 1956; see also section 5 of the entry on Naturalism in Legal Philosophy ).
Similarly, various strands in the influential American legal academic enterprise of ‘economic analysis of law’ share Bentham’s debunking attitude toward central legal terms such as ‘right’ and ‘obligation’. Like Bentham, some economic analysts oscillate (or equivocate) between (1) a moral theory that reduces those normative terms to terms describing maximization of human satisfactions, and (2) a theory that does not engage in moral argument, but only claims to describe human motivation, accounting for terms such as ‘right’ and ‘obligation’ as rhetorical epithets that agents use to pursue what they will [see the entry on The Economic Analysis of Law, section 2.2].
Not all legal sceptics have been driven by the empiricism of Bentham and the Scandinavians. Many other forms of scepticism about law have also sought support in scepticism about the meaning of language. Attacks on the coherence of the idea of the rule of law, and on the meaningfulness of legal discourse, have used ideas in the philosophy of language as diverse as Saul Kripke’s interpretation of Wittgenstein’s remarks on rule following (see entry on Interpretation and Coherence in Legal Reasoning ), and deconstruction (see Endicott 2000, chapter 1).
2. The use of language in law
2.1 Law and signs
There are insurmountable objections to Bentham’s idea that a law is an assemblage of signs. Law (in the sense that is relevant here) is the systematic regulation of the life of a community by standards treated as binding the members of the community and its institutions. A law is a standard that is part of such a systematic form of regulation. Many such standards have no canonical linguistic formulation (that is, no form of words which, according to law, determines the content of the standard). Lawyers in common law systems are familiar with such norms: murder may be a criminal offence (or slander may be a tort, or certain agreements may be enforceable as contracts…), not because of the expression by any person or institution of a rule that it should be so, but because the institutions of the legal system customarily treat murder as an offence (or slander as a tort...). Moreover, common law systems cannot be distinguished from legal systems consisting only of linguistic acts, because no legal system consists only of linguistic acts. A civil law system with a civil code and a criminal code may make murder an offence (and slander a tort…) by a written act, and it may be a written constitution that gives legal force to the civil code and to the criminal code. But the validity of the written constitution will depend on a norm which is not created by the use of signs: the rule that that text is to be treated as setting out the constitution.
Bentham and John Austin knew that there are rules of law that were not laid down in language. Preserving their view of law as signification of volition, they accounted for such laws as tacit commands of the sovereign. That convoluted device cannot provide the resources needed to explain the existence of a norm. In the right circumstances, it is certainly possible to communicate without using signs (and in particular, it is possible to convey a volition, backed by a threat of force, just by saying or writing nothing). But silence can only be a means of communication when the circumstances give it a meaning. We can say that a tacit command has been issued only if it is possible to identify features of the situation that distinguish the tacit communication from mere inaction communicating nothing [see Hart 2012, 45–48]. Those features do not generally accompany customary rules (in fact, they generally do not accompany customary rules).
There is another conclusive objection to the idea that a law is an assemblage of signs. When a lawmaking authority does use language to make a law (as when a legislature uses a lawful process to pass an enactment that is within its powers), the resulting law is not an assemblage of signs. When I write you an email message to make a request, the message is an assemblage of signs, but the request is not an assemblage of signs; it is something I use the assemblage of signs to make. The law that results from the use of language in an enactment is a standard (or standards) whose existence and content are determined by the legal effect that the law ascribes to that use of an assemblage of signs.
A law, therefore, is not an assemblage of signs, and law is not necessarily made by the use of language, and every legal system has norms that were not made by the use of language. Laws are not linguistic acts. Laws, you may say, are standards of behaviour that can be communicated (and some of which are made) by using language. But even this is controversial among writers on legal interpretation.
2.2 Language and legal interpretation
What is the relationship between the language that is used to make legal standards, and the law itself? If the law provides that a form of words determines the content of a standard (such as a term of a contract, or a criminal offence, or a duty of the executor of a will), what is the effect of the use of the words? The question seems to demand general theories of the meaning of language and of the interpretation of communicative acts. If there are no general theories to be had, then there is no general answer to the question. A theory of meaning and interpretation of legal language would not be very much less general than a theory of meaning and interpretation of language.
Legal theorists have tried to construct theories of the meaning of legal language, and theories of legal interpretation, based on specific features of law, of legal systems, and of the use of language in making law. The entries on Interpretation and Coherence in Legal Reasoning, Constitutionalism [see section 7, ‘Constitutional Interpretation’], The Nature of Law, and Naturalism in Legal Philosophy address theories of legal interpretation. Here I will address features of language that raise challenges for philosophy of law and for philosophy of language. Any good account of the meaning and interpretation of the language used in making law needs to deal with ways in which its legal effect depends on the context of its use (addressed in this section). ‘Pragmatic’ effects of language use, such as context-dependence, have long been the matter of jurisprudential debate, and recently the use of Gricean and post-Gricean pragmatics has become a major field of debate among philosophers of language and philosophers of law (section 2.3). These pragmatic features of communication raise insuperable difficulties for any attempt to give a true account of legal interpretation that is well enough organised to deserve to be called a ‘theory’. Debates over the pragmatics of legal language are often premised on the view that the effect of a lawmaking use of language is that the content of the linguistic communication in question determines the content of the law. But even this premise is controversial, and has been denied by theorists who account for the law as being determined by moral reasons for conclusions as to what rights, duties, powers, and liabilities people have, given the facts of legal practice and history (section 2.4).
As an instance of controversy over the effect of the use of language in law, consider the case of Garner v Burr  1 KB 31. The legislature had made it an offence to use a ‘vehicle’ on a road without pneumatic tires. Lawrence Burr fitted iron wheels to his chicken coop, and pulled it down the road behind his tractor. Burr was prosecuted under the statute. The magistrates acquitted him, apparently on the ground that a chicken coop is not a ‘vehicle’. The appeal court reversed that decision. The Lord Chief Justice wrote,
The regulations are designed for a variety of reasons, among them the protection of road surfaces; and, as this vehicle had ordinary iron tyres, not pneumatic tyres, it was liable to damage the roads. [The magistrates] have put what is in my opinion too narrow an interpretation on the word ‘vehicle’ for the purposes of this Act. It is true that, according to the dictionary definition, a ‘vehicle’ is primarily to be regarded as a means of conveyance provided with wheels or runners and used for the carriage of persons or goods. It is true that the [magistrates] do not find that anything was carried in the vehicle at the time; but I think that the Act is clearly aimed at anything which will run on wheels which is being drawn by a tractor or another motor vehicle. Accordingly, an offence was committed here. It follows that [the magistrates] ought to have found that this poultry shed was a vehicle within the meaning of s. 1 of the Road Traffic Act of 1930. ( 1 KB 31 at 33)
The magistrates and the appeal court seem to have disagreed over the effect of two principles: a principle that statutes should only be read as imposing criminal liability if they do so unequivocally, and a principle that the purposes for which Parliament passed the statute should be pursued. Assume that those principles are legal principles, in the sense that a decision in accordance with the law must respect them. The apparent tension between the principles might be resolved in one of the following two ways. We do not have the magistrates’ reasons, but let’s presume that they resolved the tension in the first way; the appeal court resolved it in the second way:
1. by concluding that Parliament’s purposes can be respected appropriately while still construing the prohibition strictly (so that it is no offence to pull a chicken coop down the road on iron wheels, because the chicken coop is not undeniably a vehicle), or
2. by concluding that Parliament’s purpose is sufficiently clear that it can be pursued by convicting Mr.Burr, without jeopardising the principle that criminal liabilities ought to be clearly spelled out, even if someone might reasonably claim that a chicken coop on wheels is not a ‘vehicle’.
Description of being a lawyer
It may seem that this common sort of disagreement tells us nothing about language, except perhaps that language is of no particular importance in law. It may seem that the two courts did not disagree over any question of language (after all, everyone involved was competent in English), but only over whether they ought to give effect to Parliament’s evident purpose of protecting roads by convicting Mr.Burr, or whether that would be unfair to him.
One explanation of the decision would be that the appeal court ignored the word ‘vehicle’, and treated the pneumatic tires rule as if it applied to anything that moves on wheels (instead of applying to vehicles, which is what the legislation said). But the Lord Chief Justice did not explain his decision that way. He did not hold that, because Parliament aimed to protect roads from iron wheels, Burr should be convicted regardless of the meaning of the word ‘vehicle’. He held that the magistrates ought to have found that the chicken coop was a vehicle, ‘for the purposes’ of the Road Traffic Act. Presumably the magistrates, too, considered themselves to be giving effect to the language of the Act, rather than ignoring (in the interests of fairness) what Parliament had enacted.
No doubt, legal decisionmakers sometimes depart from the language of valid enactments (or wills, or contracts...). They may do so corruptly, or in the exercise of an equitable jurisdiction to depart from the law, or because they consider that justice demands it even if the law accords them no power to depart from the language. But the magistrates and the appeal judges evidently saw themselves as giving effect to the language of the Road Traffic Act in Garner v Burr. And the sort of disagreement that arose in that case (disagreement over the legal effect of the use of a word) is so common that we seem to find a paradox: competent speakers of the English language share a knowledge of the meaning of the word ‘vehicle’, yet they disagree—apparently sincerely—over the effect of legislation using the word.
To resolve the apparent paradox, we can say that what speakers of the English language share, in virtue of their grasp of the meaning of a word like ‘vehicle’, is an ability to use the word in a way that depends on the context. The question of whether a chicken coop on wheels counts as a ‘vehicle’ would be a different question (and might have a different answer), if another statute or regulation imposed a tax on ‘vehicles’. The relevant considerations might be different again if a statute or regulation required ‘vehicles’ to keep to the left-hand side of the road. The Lord Chief Justice was right that a dictionary definition of ‘vehicle’ could not conclude the question of whether the chicken coop was a vehicle in Garner v Burr, because the purpose of a dictionary definition is to point the reader to features of the use of the word that can be more-or-less important in a variety of more-or-less analogical ways in various contexts. A definition of ‘vehicle’ as a mode of conveyance offers the reader one central strand in the use of that word, but does not tell the reader whether a more-or-less analogical extension of the word to a chicken coop on wheels is warranted or unwarranted by the meaning of the word. Another way of stating this resolution of the apparent paradox is by distinguishing between the meaning of a word (which the magistrates and the appeal judges all knew) and the way in which a communicative act using the word ought to be understood (over which they disagreed). What the judges and magistrates in Garner v Burr shared was a knowledge of the meaning of the word ‘vehicle’, and what they disagreed over was the effect of the legislature’s use of the word.
2.3 The pragmatics of legal language
The dependence of the effect of legal language on context is an instance of a general feature of communication, which some philosophers of language have approached by distinguishing semantics from pragmatics. The distinction is, roughly, between the meaning of a word or phrase or other linguistic expression, and the effect that is to be ascribed to the use of the expression in a particular way, by a particular user of the language, in a particular context. The pragmatics of legal language is a vast field, because the term ‘pragmatics’ could be used as a heading for much of what modern legal scholars and theorists have described as grounds for interpretation (and also as a heading for much that they have described as the theory of interpretation -since ‘pragmatics’ is a term not only for effects of communication, but also for the study of those effects). For example, the work of the judges in Garner v Burr can be described as an exercise in pragmatic inference. The technical sound of the word ‘pragmatics’ may suggest that it is a term for the theoretical study of its object; in fact, the field of study is what might be inferred from the fact that someone said what they said in the context in which they said it. No object of study is less apt for theorising.
This vast field has been much studied over millennia by lawyers and philosophers, but only recently through the use of resources that may be borrowed from the 20th-century development of ‘pragmatics’ as a discipline in linguistics and the philosophy of language. Much recent work in the pragmatics of legal language has focused on Gricean and post-Gricean pragmatics, as a source of insights into the theory of legal interpretation (see e.g. Marmor 2008, 2014, Soames 2008, Ekins 2012 (see pp 205-211), Carston 2013, Solum 2013; on Gricean and post-Gricean pragmatics, see Pragmatics ).
It is controversial whether legal pragmatics is simply a part of the pragmatics of language use in general (see Dascal and Wróblewski 1988). It stands to reason that if the pragmatics of language use depends on the context of an utterance, the legal context of a lawmaking use of language will have implications for the meaning conveyed and, therefore, for the law that is made. Scott Soames has argued that the question of the relation between the content of the law, and authoritative sources such as statutes, is ‘an instance of the more general question of what determines the contents of ordinary linguistic texts’ (Soames 2008). It is a popular idea that when a legislature enacts a statute, ‘The content of the statute is what the lawmakers asserted and committed themselves to, in adopting the statutory language’ (Soames 2008). Speakers can commit themselves to something that they have not asserted, and can make assertions in contexts that make it clear that they are not committed to what they have asserted.
But Andrei Marmor has argued that the pragmatics of legal language is unique in crucial respects (Marmor 2008), with pragmatic considerations playing some very significantly lesser role in law, than in other contexts: ‘In ordinary conversations, pragmatic enrichment is the norm, not the exception; in statutory law, it is the exception’ (Marmor 2014, 34). In his view, the Gricean maxims of conversation, derived from a general principle of cooperation, do not straightforwardly apply in legislative communication, because legislative communication is ‘strategic’ rather than cooperative (Marmor 2014, 45-57). Likewise, Francesca Poggi has argued that Grice’s theory does not apply to legislation, because of the ‘conflictual behaviour’ of those addressed by legislation (Poggi 2011, 35). As Marmor points out, speakers and listeners in strategic communication take advantage of uncertainties in the implicatures to be derived from pragmatic norms of communication. For present purposes, we can think of an implicature as a proposition that a person is to be understood to have conveyed by using language in a particular way, although it is not expressed by what the person said (see Grice 1975, 43-4). Speakers may exploit implicatures in order to avoid asserting what they prefer to implicate, and listeners may exploit uncertainties concerning implicatures in order to bend the effect of a communication to their liking. Yet these aspects of communication are not distinctive to law making. And they depend on cooperation among participants in the ‘conversation’ in question: some degree of cooperation is necessary in order for it to make any sense for a speaker to take advantage of an implicature, or for a listener to treat an uncertain implicature as a genuine implicature.
What has been called ‘near-side’ pragmatics is a matter of determining what is said by an utterance; in informal conversation it involves the use of indexicals and pronominal figures that are avoided in legislation, and involves the implications of context for an understanding of the uses of terms. And ‘far-side’ pragmatics (determining what is implicated rather than said, and determining what speech acts are carried out by means of an utterance) can involve figurative uses of language, sarcasm, innuendos, and other techniques that lawmakers avoid (for Korta and Perry’s distinction between ‘near-side’ and ‘far-side’ pragmatics, see Pragmatics ). There is some force in Marmor’s claim that pragmatics is different in law. Legislation systematically avoids many communicative devices that can be very useful in other contexts of communication. But those devices are also avoided in a very wide range of other formalized and technical uses of language—for example, in most wedding invitations, in Wikipedia, and in well-written instruction guides for applying for bank account or a university programme or for operating a rice cooker.
Yet the pragmatic effects of communication are essential in all those contexts of language use, and are essential to the legislative use of language. This is true both of ‘near-side’ and ‘far-side’ pragmatics. The discussion of the decision in Garner v Burr, above, provides an illustration of the ways in which the context of a legislative enactment can determine the reference of terms such as ‘vehicle’, and this determination of meaning is an aspect of near-side pragmatics. The dispute over the requirement of pneumatic tires on ‘vehicles’ in Garner v Burr can be understood as a dispute over the effect of the context on the content of the regulatory provision; another way of making the same point is to describe it as a dispute over the implicatures to be drawn from the fact of the enactment of that provision. This aspect of the pragmatics of legal language is simply an instance of the pragmatics of the use of language in general (it would be very easy to formulate a version of the apparent paradox from Garner v Burr, discussed in section 2.2 above, in a non-legal context).
The role of far-side pragmatics is illustrated by the master implicature of every legislative act: it is never said in legislation, but there is always an implicature, that the rights and duties and powers established by the legislation are to be legal rights and duties and powers, under the law of the jurisdiction for which the enactment is made (see Endicott 2014, 55). Very many other more particular lawyers’ presumptions serve as maxims of the pragmatics of lawmaking (such as the presumption that a new enactment overrides a contrary previous enactment).
It is worth noting, as well, that the diversity of contexts of communication (even the diversity of conversations that the same two people may have, let alone all the diversity of exchanges that might be called ‘conversations’ by more or less remote analogy) is so great, that the variations among pragmatic aspects of ordinary conversations are much wider and deeper, than any general distinction that can be drawn between legislative communication as a class, and ordinary conversations as a class. The use of language in making law is so closely interwoven with the use of language to exercise authority in other contexts, and with the use of language in games, in talk of what is right and wrong, and generally in communication in families and in organisations of every kind, that the pragmatics of legal language is best seen as a deeply integral part of the pragmatics of the use of language in general.
It may seem otherwise, because the persons to whom laws are addressed can be extraordinarily creative in finding ways in which they can fit the laws to their own interests, and persons in a dispute over the law tend to disagree deeply over the effect of the language that has been used to make law (so that it is not at all unusual for each side to end up with a deeply-felt conviction that the law is so clearly on his or her own side, that no reasonable person could think otherwise). So it seems, as Marmor argues, that the use of language to make law is not part of a cooperative exchange between lawmakers and the persons to whom their communications are addressed. For this very reason, of course, every legal system has tribunals for the resolution of disputes; they are always an integral part of the system of governance in a state that is ruled by law, although they very often have various forms of independence from other institutions of the system. The tribunals or courts typically have authority to interpret the language of the law, and to determine its effect in particular cases. Unlike the parties to a dispute, courts are characteristically involved with the lawmakers in the enterprise of governance (even where the courts are independent), and often have their own lawmaking capacities. Their role in the system is the substratum for a cooperative principle that has even deeper significance (and somewhat simpler implications) in the operation of a legal system, than Grice’s cooperative principle has in ordinary conversation. And, of course, that cooperative role of a court in the legal system constrains the arguments of parties to a dispute: however strongly they are inclined to twist the language of the law (and however strongly they may feel that the other side is twisting the language of the law), they need an argument that will persuade an institution that is committed, at least in principle, to cooperative interpretation of lawmaking acts.
The problem faced in cases like Garner v Burr is an especially vivid reminder of a problem that philosophers of language have long been more or less aware of (see Aristotle’s discussion of the notion of ‘friendship’ in Eudemian Ethics VII, 2, 1236a 33). The context-dependence of the meaning of utterances requires an account of linguistic competence that relates it to other human capacities—capacities to judge the importance of context and to draw analogies. It would be a mistake in the philosophy of language to account for language in a way that divorces its mastery from other aspects of reason. That connection between the use of language and reason is also illustrated by the contentious nature of the use of language in law. Much work on pragmatics by philosophers of language has concerned the conditions for understanding an utterance; judicial disputes over language are a reminder that those conditions—even though they are conditions of intelligibility of communication—lend themselves to disagreement and to reasoned dispute over how an utterance is to be understood (see Carston 2013, for discussion of controversies over the applications of conversational maxims). The special feature that distinguishes the legal use of language from ordinary conversation is not that participants in a legal system act strategically while participants in an ordinary conversation act cooperatively; the special feature is that legal systems need institutions and processes for resolution of the disputes about the application of language that arise as a result of its context-dependence, and as a result of other pragmatic aspects of communication.
2.4 The Semantic Sting
Discussions of the pragmatics of legal language are expressly or implicitly premised on a view of the relation between a law-making use of language, and the law that is made. It is the view that if a body or person is authorised to make law, it makes the law that it communicates by its use of language.
I will call that ‘the communication model’. It must be qualified in at least four ways, because the law itself regulates the making of law:
the law that is made will be limited by any limit on the power of the law maker (as to the substance of the law that it can make, or as to the process by which it can lawfully make law), and
rules of law may qualify the law that is made, in a variety of ways that are not susceptible of any general characterisation
(see Solum 2013 for an extensive discussion of ways in which the law that is made by a communicative act may be qualified), and
courts may resolve indeterminacies in the effect of an act of law making, and where they do so, their decisions may have conclusive legal effect, and
if a court departs from what the law maker communicated (for good reasons or bad), the decision of the court may have conclusive legal effect (for the parties, and also for the future if the decision is treated as a precedent).
Even with those qualifications, some theorists reject the communication model. They argue that the identification of legal rights and duties cannot be based merely on facts, such as the fact that an authority has communicated such-and-such. Mark Greenberg has argued against the communication model (Greenberg 2011, 221). In his view,
Trying to understand legislation on the model of communication is misguided because legislation and legislative systems have purposes that have no parallel in the case of communication and that may be better served if a statute’s contribution to the content of the law is not constituted by what is communicated by the legislature. (Ibid; for an argument against Greenberg’s claim, see Marmor 2014, 15-22)
The most influential theory to have denied the communication model is that of Ronald Dworkin, as it emerged in his argument against the idea that the law of a country is a system of rules. had tried to explain the nature of a legal system on the basis of social rules for identifying the law of a community (‘rules of recognition’), which provide shared tests of validity for rules of the system. Hart treated it as obvious that such rules of recognition commonly identify the acts of law-making authorities as sources of law, conferring power on those authorities to make law, and that (subject to qualifications on the communication model such as those listed above) the law becomes what a lawmaker conveys that it is to be, in the communicative act by which it exercises such a power.
On that view, the content of the law of a community depends on a conventional way of recognizing legal rules. But if members of the community share a way of recognizing their law, how can they engage in the deep disputes about the law that we are all so familiar with? Dworkin made that question into a focus of jurisprudential debate. And he framed the question as an objection to a misguided view of language, and of the relation between law and language. He began his book Law’s Empire (Dworkin 1986b) by contending that legal theorists like Hart cannot explain theoretical disagreement in legal practice, because they think that lawyers share uncontroversial tests (tests which Dworkin termed ‘criteria’) for the truth of propositions of law. The ‘semantic sting’ involves the misconception that the language of the law can be meaningful only if lawyers share such criteria. It is fatal to a legal theory, because it leads the theorist to think that people cannot have any deep (or ‘substantive’ or ‘genuine’) disagreement about the law. They can only disagree (1) about empirical questions such as what words were used in a statute, or (2) about how to resolve penumbral cases, or (3) about whether the law should be changed. If you suffer from the semantic sting, you will conclude that disagreeing about the criteria for application of the language of the law would be like using the same words with different meanings. People who disagree in that way are only talking past each other. Here is how Dworkin sets out the views of theorists who suffer from the semantic sting:
We follow shared rules, they say, in using any word: these rules set out criteria that supply the word’s meaning. Our rules for using ‘law’ tie law to plain historical fact. It does not follow that all lawyers are aware of these rules in the sense of being able to state them in some crisp and comprehensive form. For we all follow rules given by our common language of which we are not fully aware. …We all use the same factual criteria in framing, accepting, and rejecting statements about what the law is, but we are ignorant of what these criteria are. Philosophers of law must elucidate them for us by a sensitive study of how we speak. They may disagree among themselves, but that alone casts no doubt on their common assumption, which is that we do share some set of standards about how ‘law’ is to be used. (Dworkin, 1986b, 31)
Dworkin argued that legal philosophy needs to make a fresh start, in order to face the challenge of explaining disagreement about the law. Any theory of law, he claimed, needs to be a ‘constructive interpretation’ of legal practice. A constructive interpretation is one that simultaneously fits the facts of the object of interpretation (in the case of legal interpretation, the object is not some particular communicative act by a lawmaker, but the entire practice of the community) and portrays it as an object that achieves its purpose. On this view a theory that does not present law as an exercise in constructive interpretation cannot even compete as a theory of law, because it suffers from the semantic sting. And the communication model is incompatible with the idea that law is an exercise in constructive interpretation.
The semantic sting really does seem a gruesome fate for a lawyer, because those who suffer from it would have to say that no one really disagrees with anyone about the law: whenever people think they disagree about how to identify the law on any point, that fact in itself would demonstrate that there is nothing to disagree about: there is no law on the point at issue, if the agreed ways of identifying the law do not decide the point. So if you suffer from the semantic sting, you will conclude that sincere, competent lawyers will never disagree with each other. When an apparent disagreement arises, they will throw up their hands and say that there is no law on the point. If you suffer from the semantic sting, you will think that at least one party to any dispute over the content of the law is both legally inept and philosophically misguided, or is just a liar.
If you think that law is made by the communicative acts of lawmaking authorities, and if everyone understands the language in question, then it may seem that you are doomed to conclude that competent lawyers cannot disagree in good faith as to what law has been made. To meet Dworkin’s challenge, it is necessary to explain how it can be reasonable to disagree over the application of rules made by communicative acts. How can competent and (let’s presume) sincere and reasonable adjudicators differ (as the Lord Chief Justice and the magistrates differed in Garner v Burr) on the question of what the law allows or prohibits?
The potential for such a reasonable disagreement (even among people who understand the language of the lawmaker equally well) is very readily to be found in the complex pragmatic considerations discussed in section 2.3 above, concerning the effect of a lawmaking communicative act. If we accept the communication model, we will conclude that participants in legal practice typically share an understanding of the meaning of the language of legislation; yet it is still clear that there are bound to be deep disagreements as to the implications of those law-making acts in some cases. Disagreement about the content of the law will arise if the persons subject to the law must apply a test that, like the legislation in Garner v Burr, can be controversial in its application in some unusual case. Likewise, disagreement will arise if the legislative act delegates authority to a court (as legislatures often do), or if the legislature communicates unclearly. The persons subject to the law must decide (and may dispute) how to interpret the acts of the legislature. And it may also be unclear (and controversial) in some cases whether the legislature has the power that it claims to have exercised, or whether a court has power to depart (and if so, whether it would be right for it to depart) from a decision of the legislature.
Does the recent outpouring of work on the pragmatics of legal language provide a successful defence of the communication model against Dworkin’s semantic sting argument? Dworkin would presumably have extended his ‘semantic sting’ argument into an argument that the communication model suffers from a ‘semantic-and-pragmatic sting’—unless, that is, it evolves into a theory of ‘constructive interpretation’, holding that the ‘pragmatic’ aspects of the lawmaking use of language require the interpreter to construct the theory of rights and duties that best fits and justifies the whole pattern of legal practice. But Dworkin’s semantic sting argument cannot in itself undermine the communication model, because his argument was based on an impoverished view of the controversy that can result from communication.
2.5 Evaluative considerations and the nature of law
The importance of the context of the word’s use requires anyone addressing the problem in Garner v Burr to make evaluative judgments, just to apply the putatively descriptive term ‘vehicle’. The context of use is a criminal prohibition imposed for a presumably good public purpose of protecting road surfaces. To determine in that context whether the word ‘vehicle’ extends to a chicken coop on wheels, you need to address (and to resolve any tension between) the two principles mentioned above: that the purposes for which Parliament passed the statute should be pursued, and that criminal liability should not be extended to conduct that has not been unequivocally prohibited. Because of that context, the question of the meaning and application of the language of the statute cannot be answered without making judgments on normative questions of how those principles are to be respected. The role of evaluative considerations in the application of ordinary descriptive terms like ‘vehicle’ raises a challenge for any thorough-going scepticism about value: the challenge is to decide whether to adopt a thorough-going scepticism about the meaning and application of descriptive language, which seems to follow from general scepticism concerning value.
For philosophy of law, the dependence of descriptive language on evaluative considerations raises special problems. If you cannot tell whether to describe an object as a ‘vehicle’ without evaluative reasoning, then you cannot tell what law the Road Traffic Act makes merely by pointing out physical facts (such as that there were wheels on the chicken coop) and social facts (such as that Parliament used the word ‘vehicle’ in the Act, or the complex social fact of the whole pattern of uses that have been made of the word ‘vehicle’). You cannot identify the law (that is, you cannot tell what legal rights and duties people have) without evaluative reasoning. That conclusion, if it is sound, seems to have important implications for the long-running debates in theory of law over the relation between fact and value in law, and over relations between law and morality. The conclusion seems to contradict one of the most provocative and controversial claims in the theory of law—the ‘sources thesis’, which Joseph Raz has stated as follows:
All law is source-based. A law is source-based if its existence and content can be identified by reference to social facts alone, without resort to any evaluative argument. (Raz (1994), 194–5; see entry The Nature of Law )
We can only decide whether the chicken coop was a ‘vehicle’ for the purpose of the Road Traffic Act by understanding the Act as pursuing something that the legislature decided was worth pursuing, and by making an evaluative judgment as to whether due respect for the pursuit of that value through the enactment, and due respect for the requirement of certainty in criminal liability, support the conclusion that the chicken coop is a ‘vehicle’ in the relevant sense. So we cannot apply the law without resort to evaluative reasoning.
It may seem that judges need to make such evaluative judgments only in borderline cases for the application of legal language—and that they have discretion in those cases (that is, the judge must make a decision that is not determined by law). But a clear case of a vehicle counts as a ‘vehicle’ for the purpose of the Road Traffic Act just because the evaluative considerations that justify the use of the word ‘vehicle’ in that context clearly support its application. The sources thesis seems to be contradicted even in the clearest cases of the application of a law stated in descriptive language, if the content of such laws can only be identified on the basis of an evaluative judgment as to how the purposes of the law ought to be conceived. That form of evaluation, you may say, can only be carried out by engaging in the very same form of reasoning that, in Raz’s theory, law excludes.
It is true that, in order to decide what the sources have directed (and thereby, in Raz’s terms, to identify the existence and content of a law), you need to understand the sense in which a word like ‘vehicle’ is used. But the existence and content of the offence can still be identified without first judging whether it ought to be an offence to do what Mr.Burr did, or whether there ought to be any offence at all of driving without pneumatic tires. The sources thesis articulates this important insight: in English law, there was no offence of driving without pneumatic tires (whether or not it would be a good idea for that to be an offence), until Parliament acted to create it. And then, it became an offence because Parliament had done so (whether or not it was a good idea to do so). Moreover, if the institutions of the law had not been prepared to treat the chicken coop as a vehicle, then because of that social fact, it would have been false (after a decision with precedential effect) to say that it was an offence to pull a chicken coop on iron wheels on the road. Because law is systematic (in the sense that the law gives legal institutions authority to identify the law), the courts’ decisions determine legal rights and obligations. So, we might say, a law is source-based if its existence and content can be identified by reference to social facts alone, without resort to any evaluative argument, other than whatever evaluative argument may be needed to determine the social facts, or to work out their implications for the case at hand.
Raz’s explanation of the nature of law is not undermined by the fact that evaluative judgments are necessary in order to identify the implications of the social fact that Parliament prohibited vehicles from being driven without pneumatic tires. It is still possible for legal directives to have the exclusionary force that, in his theory of authority, they claim. The court in Garner v Burr can decide whether it is an offence to pull a chicken coop on wheels without pneumatic tires, without answering the question, ‘should it be an offence to drag a chicken coop down a road on iron wheels?’, or the question, ‘should it be an offence to use a vehicle without pneumatic tires?’ But the court needs to ask the related question: ‘what is the relevant sense of “vehicle” for the purposes of this Act?’ It is what Parliament did that determines Lawrence Burr’s liability; it may take evaluative reasoning to answer the question of social fact, ‘what did Parliament prohibit, when it prohibited vehicles from being driven without pneumatic tires?’
2.6 Vagueness in language and in law
A speed limit on a highway is a fairly precise law: in most cases it is clear whether a driver has conformed to the standard. But highway traffic regulation also needs (and typically uses) rules against careless or dangerous driving. Such abstract standards are designed to control a variety of behaviour that lacks the uniform, measurable feature (speed) that allows the precision of a speed limit.
Vague laws, such as a rule against careless driving, pose problems for philosophy of law that are related to problems that philosophers of language and of logic have addressed in arguments about the paradox of the heap (see the entry on the Sorites Paradox ). Suppose that, according to law, it counts as careless driving to drive with bald tires. If the law gives a precise definition of the thickness of tire tread that counts as bald, then in this regard the law is (more or less) precise, and for the purposes of the law, virtually every tire is either clearly bald or clearly not bald. But if there is no such precise standard, then there are ‘borderline cases’ in which it is neither clearly true that a tire is bald, nor clearly false. And we can construct a sorites series, and a sorites paradox for the application of the law:
If a tire is not bald, it does not become bald by losing one molecule of rubber from its tread.
So a tire that has lost one molecule of rubber is not bald.
So a tire that has lost two molecules of rubber is not bald.
N. So a tire never goes bald.
N+1. So no one can ever break the rule against careless driving by driving with bald tires.
The false conclusions N and N+1 arise by apparently valid reasoning from apparently true premises. Philosophical approaches to the 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.
‘Epistemic’ arguments that the only way to solve the paradox is to deny the truth of step 2 (so that vagueness is a problem of ignorance as to where the sharp boundary is between tires that are and are not bald) imply that there is always a single right answer to the application of a law stated in vague language. Arguments that the application of a vague expression is indeterminate in borderline cases (or in some borderline cases) imply that the application of a law that can be expressed in vague language is indeterminate in some cases. But it is even controversial whether such theories matter to legal philosophy [Schiffer, 2001 and Greenawalt, 2001].
Philosophers of law have not been especially concerned with the question of how to solve (or to resolve) the paradox, but they have debated the nature of borderline cases, and its implications for the role of judges in a community, and 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 decide the matter by applying the law, since the law does not determine the matter (and in fact, no one can use the law to guide their conduct).
Some legal philosophers have responded to this problem by claiming that judges never (or virtually never) have such a choice, and that there is virtually always a right answer to a question of legal rights [Dworkin, 1986a, 1991]. Others have responded to the problem by claiming that the law gives judges discretion, in all or some borderline cases, to decide issues that the law does not determine [Hart, 2012, chapter VII.1]. That is, the standards of the system leave a choice to judges as to how to decide the issue. Then judges must treat the parties to litigation as having liabilities or obligations or entitlements that were not determinately theirs at the time when the dispute arose. That power of judges appears to some to run contrary to the principle of the rule of law, that laws (or at least, that legal burdens) should not be imposed retroactively.
As a result, vagueness raises an extraordinarily important set of problems for the philosophy of law. It may seem to be a different set of problems from those raised by vague language in general, because of three ways in which the legal use of vague language is distinctive. First, legal systems need to regulate the effects of the vagueness of language, providing for authoritative resolution of disputes as to the effect of vague language (where in some other domains of the use of language, we can get by without any such resolution).
Secondly, vagueness in legal language is distinctive because lawmakers avoid ordinary vague expressions like ‘driving very fast’ and prefer precise speed limits (or blood alcohol limits, or tire tread measurements). When the law uses vague language, it uses abstract evaluative expressions. Lawmakers typically do not prohibit driving with ‘bald’ tires; they either impose precise measures, or address the problem as part of an abstract, general prohibition on, e.g., ‘careless’ driving. An abstract standard calls upon its subjects to construct a view of the care that, in their law, a driver owes to another person—(and not merely to ask the seemingly pointless question, how bald is a bald tire?). Such standards are a very common and a very important part of lawmaking technique. A negligence standard may require ‘reasonable care’; a constitution may define a procedural right as a right to ‘due process’, or a contract may require the delivery of goods in ‘satisfactory condition’. Those abstract terms are very different from the vague descriptive terms that philosophers of logic use to illustrate their arguments about the sorites paradox (‘heap’, ‘thin’, ‘bald’, ‘red’…). It misses the point, you might think, to say that abstract standards do not draw sharp lines, because they are not designed to draw lines at all. By using an abstract standard, the lawmaker requires the people who must apply the law to construct a theory of the standard (of care, or of process, or of condition), which will draw any line that is needed. Ronald Dworkin has claimed that abstract expressions are not vague at all—that they have a different semantics from that of vague words like ‘heap’ (Dworkin, 1986b, 17).
Thirdly, the legal context of an expression may be very different from other contexts of its use. Principles of legal interpretation (for example, a legal requirement that vague criminal enactments are to be interpreted as applying only in clear cases) may make the legal effect of the use of a word more precise than its ordinary effect. The interpretation of a legal prohibition on careless driving demands an understanding of what counts as careless for the purposes of the law.
Lawyer career outlook
Given these distinctive features of law, it may seem that logicians’ discussions of heaps and baldness are beside the point, and that legal reasoning is exempt from any implications that the paradox may have for logic or for semantics. That would be a mistake. Statements of law that use abstract expressions certainly are vague in the philosophers’ sense: that is, they appear to be susceptible to sorites reasoning. ‘Careless driving’ in its legal sense is a much more complex concept than ‘driving with bald tires’, and driving with bald tires is an instance of careless driving only in virtue of the evaluative and contextual considerations that must be understood if the law of careless driving is to be understood. Yet ‘careless driving’ is susceptible to sorites reasoning precisely because those evaluative and contextual considerations give reason to conclude that it is careless to drive with bald tires. For all their complexity (and their relation to deep questions of the rights of the citizen and the responsibilities and the liabilities that a community can justifiably impose on the citizen), those considerations do not determine a standard that is more precise than a standard that could be expressed using such an ordinary vague expression as ‘bald’. That is, those considerations do not provide a way of distinguishing between one tire in the sorites series and the next. Similarly, it is possible to construct sorites series for the application of even more abstract legal standards such as the right to due process, or the United States constitutional prohibition on cruel and unusual punishment. These evaluative tests have special legal meanings that can only be understood in the context of a legal system, and in the more particular context of a specific area of law, and by reference to their elaboration and development in a common law system of precedent. Even with a mastery of all those aspects of the context, however, the best theory of the effect of such standards will not provide a way of distinguishing between one tire in the sorites series and the next. A theory of careless driving in some particular legal system at a particular time would be a defective theory, if it yielded the conclusion that the loss of one molecule of rubber from a tire can make the difference between driving that is careless and driving that is not careless according to law.
Consider again the case of Garner v Burr (above, section 2.2). The term ‘vehicle’, as properly understood for the purposes of the Road Traffic Act, is vague if there are borderline cases for its application (cases in which it is unclear whether the term applies to some object). It may seem that nothing is more patently a borderline case of a ‘vehicle’ than a chicken coop on wheels. But we should bear in mind that (if the discussion in section 2.2 above is sound) the correct application of the term depends on legal principles relating to the purposes of the legislation and to the need for clarity in criminal liabilities. So the chicken coop on wheels is a borderline case if it is unclear whether those principles required the provision to be applied to it. And the application of the term was indeterminate in the case, if those principles did not require a decision one way or the other. Of course, the appeal court held that a conviction was required—which we might say amounts to a decision that the term applied to the chicken coop on wheels. The magistrates and the appeal court disagreed about it, but that disagreement in itself does not mean either that the application of the term was determinate or that it was indeterminate. And the fact that the appeal judges seem to have thought that the term clearly applied to the chicken coop cannot tell us that the application of the term was determinate.
The magistrates’ approach put a special emphasis on the principle of certainty in criminal liability, and the appeal court’s approach put a special emphasis on the effective accomplishment of the purpose the legislation was designed to pursue. Each approach would take the law in a different direction. The application of the term was indeterminate (before the decision of the appeal court set a precedent) if neither approach was demanded by the complex resources of legal reasoning that the courts (as a matter of legal obligation) had to act upon. If that was the case, then the appeal court’s decision was not contrary to the law. But a decision upholding the acquittal would not have been contrary to the law, either. The law as it was at the time of the decision allowed the courts to move it in either direction.
That account of a case like Garner v Burr would support the controversial claim that judges have widespread discretion in resolving legal disputes. Legal philosophers have debated whether that claim undermines the ideal of the rule of law, or reflects a basic requirement of the rule of law: that a legal system needs techniques for the resolution of legal issues that are not determined by the law (see Endicott 2000, chapter 9).
3. The use of philosophy of language in philosophy of law
3.1 Introduction: Definition as a methodology in philosophy of law
Bentham presented his theory as a definition. In focusing their attention on the meaning of the main terms of legal discourse, Bentham and John Austin were ahead of their time. But it has come to be a commonplace view in legal theory (though still controversial) that they were misguided in attempting to define those terms. So, for example, Hart in the 1960s rejected definition as useful in philosophy of law (Hart 2012, 14–17). Dworkin in the 1980s accused Hart of only repackaging the same approach as the ‘more candidly definitional’ method of John Austin (Dworkin 1986, 32–33). And Richard Posner in the 1990s accused both Hart and Dworkin of pointlessly “trying to define ‘law’” (Posner 1996, vii).
There is no reason to describe the work of Hart or Dworkin as defining the word ‘law’. And defining that word would not solve any of the problems of jurisprudence (as Hart pointed out). The fundamental reason is that a definition is useful only to someone who needs to learn the meaning of a word, and legal philosophers know the meaning of the word ‘law’. Their problems and their disputes would not be resolved by a statement that would help someone who did not know what the word meant. Philosophers of law cannot solve their problems by giving a definition of the word ‘law’, any more than philosophers of language can solve their problems by giving a definition of the word ‘language’ (for discussion of the implications for legal philosophy of the semantics and metasemantics of the word ‘law’, see Coleman and Simchen 2003).
A further reason is that, as John Finnis and Ronald Dworkin have both explained in different ways, the word ‘law’ can be used in a variety of senses: the law of the jungle, the law of gravity, laws of thought, Murphy’s law, etc. (Finnis 1980, 6; Dworkin 1986, 104). A definition would have to allow for those senses. It might be an intriguing (and arduous) study in culture and human thought to explain the analogies among those senses, but it is a study that holds out no special promise for understanding the law of a community.
3.2 Language and the normativity of law
Legal philosophers have tried to explain the normativity of law—the fact that the law of a community is, or presents itself as, a guide to the conduct of members of a community. One easy way to express this abstract feature of law is by pointing out that the law can be stated by making normative statements (i.e. statements that use expressions like ‘obligation’, ‘right’, ‘must’, ‘may’). And one attractive way of trying to explain the normativity of law is by explaining the meaning and use of the normative language that is often used in stating the law. That is, the problem of explaining the nature of legal norms (obligations, rights, etc.) can be addressed by explaining the meaning or the use of the normative words that are used in law (‘obligation’, ‘right’,…). Joseph Raz has put it that ‘The problem of the normativity of law is the problem of explaining the use of normative language in describing the law or legal situations.’ (1990, 170)
We saw that Bentham’s theory of normative language in general was that, because there is no perceptible substance or emotion for which they stand as a name, words such as ‘right’ must be ‘expounded’ by paraphrasing sentences containing them. The paraphrase would use sentences that contain only words that can be defined as referring to substances and perceptions. When no such paraphrase is available, he considered that normative language is meaningless. So he held that the phrase ‘natural rights’ is “simple nonsense: natural and imprescriptible rights, rhetorical nonsense; nonsense upon stilts” (Bentham 1843, Art.II). While the language is meaningless, he thought, we can explain its use—as a way of doing something. Using such nonsensical expressions is a way in which the speaker expresses his preferences.
Although Bentham and John Austin thought that the notion of a natural right was nonsense, they did not at all think the same about the notion of legal rights. In their command theory they found a way of paraphrasing normative expressions in their legal use. They explained the meaning of the word ‘duty’—and thereby the normativity of law—by reference to the pain and pleasure that a superior offered as motivation for compliance with the superior’s will:
That is my duty to do, which I am liable to be punished, according to law, if I do not do: this is the original, ordinary, and proper sense of the word duty. (Bentham 1776, 109; cf. John Austin 1832, 14).
Hart used the resources of twentieth-century philosophy to challenge that approach to normative language. He drew on the work of J.L. Austin, a philosopher of language who thought that problems in many areas of philosophy could be dissolved by pointing out the things that people do with words. J.L. Austin once suggested that “a statement of ‘the law’” is a performative statement, rather than “a statement of fact” (Austin 1962, 4 n.2). The suggestion is that to state the law is to perform an act (an act other than the making of an assertion that could be true or false). J.L. Austin’s claim offered to account for the normativity of law by reference to the things that people do with words. That hint attracted Hart (Hart, 1954), whose theory of law is based on a ‘practice theory of rules’. He articulated the theory by pointing out what people do with the normative language they use in stating rules.
Hart started by arguing that Bentham and Austin had explained the meaning and use of normative language in a way that failed to account for its role in ordinary discourse. He pointed out that their account of the meaning of the word ‘duty’ left them unable to draw a distinction that people ordinarily draw in their use of language, between the command of a gunman (which no one would say imposes a ‘duty’ or an ‘obligation’), and a legal prescription:
The plausibility of the claim that the gunman situation displays the meaning of obligation lies in the fact that it is certainly one in which we would say that B, if he obeyed, was ‘obliged’ to hand over his money. It is, however, equally certain that we should misdescribe the situation if we said, on these facts, that B ‘had an obligation’ or a ‘duty’ to hand over the money. So from the start it is evident that we need something else for an understanding of the idea of obligation. There is a difference, yet to be explained, between the assertion that someone was obliged to do something and the assertion that he had an obligation to do it. (Hart 2012, 82)
Bentham and Austin would have had a ready response: that people ordinarily misuse the word ‘obligation’. They fail to give it a meaning that can be expounded by reference to sensible objects. Bentham and Austin were linguistic philosophers, but (unlike J.L. Austin) not ordinary language philosophers. They sought a way of using language that would back up their empiricism and utilitarianism, and they were actually pleased if that technique called for a reorganisation of ordinary language: it showed that they were disclosing what had been obscured by prejudice and cant.
Hart’s approach to language was different. It may seem that he did not need to talk about language, and that his point could have been made without mentioning types of assertions, or what ‘we would say’. His argument, you might think, simply puts into the linguistic mode an argument that could be made with no mention of language: an argument that you can have an obligation without being liable to sanction. And yet, the linguistic form of the argument was important to Hart. He wanted to avoid explaining the difference between obligation and coercion in the way a natural law theorist might (by saying that an obligation is a kind of reason). So his focus on the use of the word ‘obligation’ is no accident. He did not point out the way we use that word as an oblique way of appealing to our shared wisdom as to what obligation is. It was actually important to him to point out how we use the word. His explanation of the normativity of law relies on the use of such words to display an attitude.
Hart claimed that a legal system is a system of power-conferring and duty-imposing rules, which are validated by a ‘rule of recognition’. That rule is not made valid by another rule; it is a ‘social rule’. It is in Hart’s explanation of that crucial notion of a social rule, that he turned to the use of words to explain the normativity of law. He claimed that a social rule is a regular pattern of conduct accompanied by a ‘distinctive normative attitude’, which ‘consists in the standing disposition of individuals to take such patterns of conduct both as guides to their own future conduct and as standards of criticism’ (Hart, 2012, 255). In accounting for that disposition, or ‘internal attitude’, Hart’s emphasis was on speech acts—on the use that participants in the practice make of normative language.
What is necessary [for a social rule to exist] is that there should be a critical reflective attitude to certain patterns of behaviour as a common standard, and that this should display itself in criticism (including self-criticism), demands for conformity, and in acknowledgments that such criticism and demands are justified, all of which find their characteristic expression in the normative terminology of ‘ought’, ‘must’, and ‘should’, ‘right’ and ‘wrong’ (Hart, 2012, 57).
Hart’s interest in normative language was focused not on its meaning, but on the attitude that people display when they use it. His explanation of the difference between non-normative and normative assertions (between, as he put it, ‘the assertion that someone was obliged to do something and the assertion that he had an obligation to do it’) was merely that the latter sort of assertion is used to display a distinctive sort of attitude.
Just as Bentham’s approach to normative language was allied to his empiricism and his utilitarianism, Hart’s approach to normative language was allied to his philosophical methodology, and to his views on the relation between law and morality. His methodological purpose was to describe human practices, and he wanted to defend a conceptual separation between law and morality—a distinction in kind between legal obligation and moral obligation (see Hart 2012, 239–240). The practice theory of rules attracted him as a tool for those purposes, because it offered a way of accounting for the normativity of law by pointing to forms of behaviour which could be described, and which did not (in his view) carry any moral baggage. In saying that people use normative language to display an attitude to regularities of behaviour, the theorist did not need to do moral philosophy, and did not need to make any moral evaluations of the practice being described. The theorist did not even need to ascribe any moral evaluations to the participants in the practice, since the fact that people use normative language to display an attitude is, Hart considered, consistent with their having various moral views or none at all.
How much simpler it would be to say that normative language means the same in its moral uses and in its legal uses. While a person who says that you must stop at a red light may display various attitudes, what he or she asserts is a necessity that implies a (presumptively) conclusive reason to stop. If the statement is a statement of law, the necessity is a legal necessity; if the statement is a moral statement, the necessity is a moral statement.
That approach is not only simpler; it closes the gap that Hart left in his theory. According to Hart, the meaning of normative language differs in morality and in law. But in fact, Hart had nothing to say about the meaning of normative expressions such as ‘ought’ and ‘must’ or ‘obligation’ or ‘right’ (except that their meaning differs in law and in morality). He only pointed out that people display an attitude when they use such language.
Joseph Raz rejected Hart’s claim that normative terms have a distinctive meaning in statements of law. In Raz’s view, to make a normative statement such as ‘you must stop at a red light’ is to imply that you have a certain sort of reason for action. If it is a statement applying the law, it implies that you have reason to stop from the point of view of the law. Raz’s theory of law, unlike Hart’s, is part of a theory of practical reason in general, and his account of normative statements treats them as having the same meaning in law and in morality. Hart initially thought that that approach would necessarily lead to an extreme sort of natural law theory, in which every true statement of law is necessarily a true moral statement, and every valid legal obligation is necessarily a moral obligation. But Raz resolved that concern of Hart’s, by pointing out that normative statements can be made in a detached way. People can make them without endorsing the point of view from which the reasons they are stating are valid (see Raz 1990, 175–177). Still, Hart did not accept Raz’s approach, which he felt created a certain sort of conceptual connection between law and morality. In responding to Raz, he insisted that ‘statements of the subject’s legal duties need have nothing directly to do with the subject’s reasons for action’ (Hart 1982, 267). And in an interview with the Spanish journal Doxa near the end of his career he maintained ‘that legal and moral obligation are conceptually different’ in the sense that a statement of obligation has a different meaning in law and in morality (De Paramo 1988). He never explained the difference in meaning; its necessity only follows from a methodological preconception.