The Semantic Sting

We saw that Hart tried to base the normative force of law on rules of recognition, which provide shared tests of validity for rules of the system. A rule of recognition is a social rule, consisting in a combination of two facts: the fact that the officials of the community regularly use certain tests to identify valid legal rules, and the fact that those officials take a ‘critical reflective attitude’ towards that behaviour. On that view, the content of the law of a community depends on a convention (or set of conventions), a way of recognizing legal rules that members of the community share. But if members of the community share a way of recognizing their law, how can they engage in the disputes about the law that we are all familiar with? How can people disagree about what the law is, if they share a conventional way of recognizing the law?

It is Ronald Dworkin who has made that question into a focus of jurisprudential debate. And he has framed the question as an objection to a misguided view of language, and of the relation between law and language. He begins his book Law's Empire by contending that many other theorists (including Hart) suffer from a ‘semantic sting’: they ‘insist that lawyers all follow certain linguistic criteria for judging propositions of law’ (Dworkin 1986, 32, 45). Dworkin claims that legal theories like Hart's cannot explain theoretical disagreement in legal practice, because they think that lawyers share uncontroversial tests provided by the conventional meaning of the word ‘law’ (tests which Dworkin terms ‘criteria’) for the truth of propositions of law. The semantic sting is 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 about empirical questions (such as what words were used in a statute), or about how penumbral cases should be resolved, or about whether the law should be changed. 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:

They say that theoretical disagreement about the grounds of law must be a pretense because the very meaning of the word ‘law’ makes law depend on certain specific criteria, and that any lawyer who rejected or challenged those criteria would he speaking self-contradictory nonsense.

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.

Philosophers who insist that lawyers all follow certain linguistic criteria for judging propositions of law have produced theories identifying these criteria. I shall call these theories collectively semantic theories of law. …

Semantic theories suppose that lawyers and judges use mainly the same criteria (though these are hidden and unrecognized) in deciding when propositions of law are true or false; they suppose that lawyers actually agree about the grounds of law.… (Dworkin, 1986b, 31–33)

Dworkin does not claim that we never share criteria for the application of a word: he says that there may be such shared, uncontroversial tests for the application of a word like ‘book’. But he claims that words like ‘law’ (and presumably most legal terms) stand for ‘interpretive concepts’. The correct application of an interpretive concept is determined not by a shared test of applicability, but by the theory that gives the best interpretation of the practice in which the concept is used. The semantic sting argument is a claim that Hart applies criterial semantics generally to legal (and, incidentally, jurisprudential) concepts, in a way that makes real disagreement about the law impossible.

Dworkin's idea that the concept of law is an interpretive concept is part of the underpinnings of his theory of law: he claims that legal philosophy needs to make a fresh start to face the challenge of explaining disagreement about the law, and he presents his theory as the best way of meeting that challenge. Any theory of law, he claims, needs to be a ‘constructive interpretation’ of legal practice. A constructive interpretation is one which simultaneously fits the facts of the practice (or other object of interpretation) and portrays it as a practice 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.

The semantic sting really does seem a gruesome fate for a lawyer, because those who suffer from it 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 demonstrates 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.

Objections to Dworkin's diagnosis of Hart's work as suffering from a semantic sting are simple and compelling: Hart never said that people share complete and uncontroversial tests for the application of the language of the law, or specifically of the word ‘law’, and to say such a thing would have thwarted his attempt to take advantage of Wittgenstein's Philosophical Investigations (Hart, 1994, 280, 297). Hart denied that he ever suffered from the semantic sting (Hart 1994, 246). Although he did claim that legal systems are based on rules of recognition, he did not say that those rules are linguistic rules, or that his theory is true because it states linguistic rules for the use of the word ‘law’. Neither he nor any other legal theorist has said that propositions of law are true if and only if they meet the uncontroversial tests of validity that people share in virtue of their knowledge of the meaning of the word ‘law’.

Some legal theorists have held that there is no genuine disagreement about law, but they are theorists who are much more sceptical about the whole enterprise than Hart was. Perhaps the semantic sting afflicts only those sceptics who think that it is impossible to agree or disagree about the law, and that it would be possible to agree about the law only if the conventional meaning of the word ‘law’ gave us shared uncontroversial tests for the content of the law. So the semantic sting argument does not undermine a theory of law like Hart's. What it does, however, is to pose in an original and striking fashion a challenge that Dworkin has raised for any legal theory, and a challenge that Hart never took up: the challenge of explaining disagreement about the content of the law. If an elucidation of the concept of law yields the insight that every legal system has rules of recognition, we might ask how it is possible for people to disagree about the content of the law. Hart's official answer was that a rule of recognition need not be very complete; like other rules, it can be vague (Hart, 1994, 147–154, 251). To Dworkin that is no answer: if the task of a rule of recognition is to provide a way of identifying the law, the tests provided need to be complete and uncontroversial, or there is no shared way of identifying the law.

But Hart did not take so high a view of his own theory's need to provide a certain and complete answer to all questions of law. Suppose that a rule of recognition identifies a legislature as having law-making power. That, to Hart, was enough to expect of a rule of recognition. It is a very common sort of rule of recognition, and it may leave many unanswered questions (as to how to interpret an act, and as to the limits of the power). But in Dworkin's view, it is no rule of recognition at all. And there is no rule of recognition in such a community unless (at least) (i) the members of the community also share uncontroversial techniques of legal interpretation that answer all questions of how an act is to be interpreted, and (ii) any limits on the power of the legislature are uncontroversial. Since many legal systems have legislatures whose powers are not clearly delimited, and since no community has an uncontroversial shared set of canons of statutory interpretation that answer all questions of interpretation, it is quite plausible to say that no legal system has a rule of recognition on Dworkin's understanding of such a rule. If that is the best understanding of a rule of recognition, then Dworkin's argument succeeds against Hart's theory of law. It succeeds not because Hart suffered from a semantic sting (there is still no reason to attribute to his theory the view that rules of recognition exist because they are required by linguistic criteria for use of the word ‘law’). It succeeds because it shows that Hart's theory cannot explain what makes a statement of law true or false.

If, on the other hand, it is right to characterize an ordinary rule granting lawmaking power as a social convention that identifies the law of the community, then Dworkin's argument fails. Hart can explain disagreement about the content of the law, by saying that the rule of recognition may require the participants in legal practice to apply a test that is very controversial: they must decide (and may dispute) how to interpret the acts of the legislature, and it may be unclear (and controversial) in some cases whether the legislature has the power that it claims to have exercised. If such a practice can be well described as a rule-governed practice for the identification of valid laws, then Hart has established his basic claim that there are rules of recognition.

But on Dworkin's view, conceiving of a rule of recognition as a very incomplete way of answering questions of law would still not explain disagreement: it would simply mean that there is no answer to most legal questions. So in order to meet Dworkin's challenge, Hart would have to show not only that rules of recognition (and other rules) may be vague, but that it can be reasonable to disagree over the application of such rules.

Think of a disagreement of the kind that divided the magistrates and the appeal court in Garner v Burr [above, section 2.2]. It would not be enough for Hart to say that there is no determinate law on the question of whether Lawrence Burr's chicken coop counted as a 'vehicle' for the purpose of the Road Traffic Act. To meet Dworkin's challenge, it would be necessary to explain how competent and (let's presume) sincere and reasonable adjudicators can differ (as they often do) on the question of what the law allows or prohibits.

That fact about Dworkin's challenge shows that there is a connection between his semantic sting argument and his view that there is a single right answer to virtually every legal dispute. Someone who suffers from the semantic sting is bound to think, on false grounds, that there are indeterminacies in the law when people disagree in applying the language of the law. Dworkin's argument makes room for his single right answer thesis. It concludes that in any unclear case of the application of an interpretive concept, the question of whether the concept applies or not is to be answered by arguments concerning the view that best fits and justifies the object of interpretation. The question cannot be sidestepped by a peremptory conclusion that the question has no answer because of the vagueness of the concept. It should be noted, however, that even if it is successful, the semantic sting argument does not give reason to accept Dworkin's right answer thesis. One mistaken way to disagree with the right answer thesis would be to think that there is no right answer to a question of law unless uncontroversial shared tests resolve the question. Avoiding that mistake does not mean concluding that there is a single right answer to every question of law.