When people talk about the law what, if anything, are they talking about? What do their sentences mean? Much of the philosophy of law has revolved around this question. In this essay, Finlay and Plunkett offer a novel answer—and a plausible reading of the answer proposed by the foremost Anglophone philosopher of law, H.L.A. Hart.
In The Concept of Law, Hart draws a distinction between external and internal legal statements. External legal statements describe people’s beliefs, attitudes, and behavior concerning legal standards. An example is “In England, they accept as law what the Queen in Parliament commands.” Internal legal statements (ILSs), by contrast, apply legal standards and usually put their conclusions using normative language.1 An example is “You (legally) ought not drive over 55 m.p.h.”.
The semantics of external legal statements is unproblematic. But ILSs are a challenge for legal positivists, like Hart, who hold that the existence and content of the law ultimately depend solely upon non-normative social facts about a community—in particular, facts about the beliefs, attitudes, and behavior of the community’s officials. If positivism is correct, why aren’t all legal statements external? In using normative language, ILSs appear to give support for non-positivist (particularly natural law) approaches. ILSs seem to describe normative facts, that is, facts that non-contingently give one reasons for action.
Positivists often respond to this problem by borrowing from semantic accounts of moral statements that deny that these statements describe normative facts. If moral statements can employ normative terms without describing normative facts, the same thing can be true of ILSs. The most commonly borrowed account is expressivism.
Consider the sentence “Hurrah John!” In uttering this sentence, one does not attribute a property (say, hurrahness) to John. If one did, the sentence would be either true or false, depending upon whether John had that property. But “Hooray John!,” although meaningful, is neither true nor false. Rather than describing John, it expresses the speaker’s positive attitude toward John. Its meaning can be understood in term of this expressive role.
Many expressivists understand the meaning of the normative terms in moral statements (like “good”) similarly. These terms express the speaker’s affective states, rather than attribute normative properties to the world.2 An advantage of this approach is that it can explain in an acceptably naturalistic fashion why someone who makes a moral statement thereby takes herself to have a reason for action: why someone who says “Keeping one’s promises is good” takes herself to have a reason to keep her promises. Her reason is the affective state expressed, not the presence of some mysterious normative state of affairs.
The prevailing reading of Hart is that he employed expressivism in his account of ILSs.3 An ILS’s meaning is tied to its role in expressing the speaker’s affective state, in particular, her acceptance of the foundational rule of the relevant legal system (which Hart calls its rule of recognition).4
But expressivism has a serious difficulty, commonly known as the Frege-Geach problem. If the meaning of “good” is tied to its role in expressing the speaker’s positive attitude, how are we to understood its meaning in non-assertoric contexts—for example, when one asks, “Is keeping promises good?,” or says,“ If keeping promises is good, then it is good for Joe to keep his promises.” If expressivism were correct, these sentences should be nonsense, the way “Is it the case that Hurrah John!?” or “If Hurrah John!, then…” are. Expressivists have tried to solve this problem, but whether they have succeeded is questionable.
The same point applies to ILSs. If the meaning of “ought” in “You (legally) ought not drive over 55 m.p.h.” is tied to its expressing the speaker’s acceptance of the relevant rule of recognition, how are we to understand its meaning in the sentences, “Is it the case that one (legally) ought not drive over 55 m.p.h.?,” or “If one (legally) ought not drive over 55 m.p.h., then Joe (legally) ought not drive 60 m.p.h.?” Clearly these sentences aren’t nonsense.
The view that normative terms describe normative facts, by contrast, does not suffer from the Frege-Geach problem. Unlike expressive sentences, descriptive sentences (such as “The cat is on the mat”) are generally understood as having two elements to their meaning: their propositional content (that-the-cat-is-on-the mat) and the way that the content put forward or used. In “The cat is on the mat” the proposition is put forward as an assertion. But in other sentences the same propositional content can be put forward non-assertorically.
Recently, philosophers working in metaethics have offered positions that share with expressivism the view the moral statements do not describe normative facts, but deny that the semantics of such statements is expressivist. The relationship between making a moral statement and having a particular affective state is pragmatic rather than being tied to the statement’s meaning. Stephen Finlay’s quasi-expressivism, which Finlay and Plunkett apply to ILSs in this paper, is an example.
We are all familiar with the use of “ought” in a hypothetical imperative—such as when one says to the failed murderer, “You ought to have used poison, rather than a gun.” Such statements can be understood as describing a probabilistic relation between an act (using poison) and the realization of an assumed end (killing someone). Philosophers commonly claim that this use of “ought” is semantically distinguishable from its use in moral contexts. Moral norms impose categorical imperatives—they provide us with reasons for action that exist whatever our inclinations. It is these categorical norms to which the term “ought” in moral statements refers. But Finlay adopts the surprising position that the semantics is the same in both contexts. A moral statement—such as “You ought to keep your promises”—describes a probabilistic relationship to an assumed end.
This isn’t to say that there is no connection between moral statements and ends. But Finlay argues that the connection is pragmatic, rather than following from the statements’ semantics. Under normal circumstances, speaking of a probabilistic relationship to an end would make a helpful contribution to a communicative exchange only if the speaker and hearer favor the end. By uttering the sentence about the probabilistic relation, the speaker will therefore pragmatically express that she favors that end—which explains why someone making a moral statement takes herself to have a reason for action. But this expression is not part of the meaning of the speaker’s moral statement: she can appropriately use the sentence without favoring the end at all.
And what if it is clear that the listener does not favor the end? What if one says “You ought to keep your promises” to someone who manifestly does not care about keeping his promises? It is here that most philosophers would say that the moral statement must be describing normative facts. That is the only way to explain the statement’s categorical nature—its applicability to the listener independent of her inclination. But Finlay again relies on pragmatics. He argues that to speak as if an end is accepted by one’s listener when it clearly is not is a rhetorical way of making a demand that it be accepted. (Something analogous goes on when one says to one’s child, “You will take out the trash.”) Semantically, the statement remains a description of a probabilistic relation to an assumed end.
I am quite skeptical about whether quasi-expressivism is an adequate account of moral statements. But it is very promising when applied to ILSs, as Finlay and Plunkett do in this essay. Under their account, ILSs state that a certain relationship to the relevant rule of recognition exists. For example, when one says, “It is the law that one ought to drive 55 m.p.h. or less,” one asserts a proposition that the rule that one ought to drive 55 m.p.h. or less satisfies the rule of recognition of the relevant legal system (say, California). That proposition is either true or false as a matter of descriptive fact.5 Because ILSs have this propositional content, the same content can occur in non-assertoric contexts. The Frege-Geach problem is solved.
Notice that although ILSs are descriptive, there remains a relationship between making an ILS and taking oneself to have a reason for action. But the relationship is the result of pragmatics, rather than following from the ILS’s semantics, as the expressivist claims. This pragmatic approach has an unexpected payoff: it can easily explain the many circumstances in which someone who makes an ILS lacks any positive attitude toward the rule of recognition. Many ILSs are, as Joseph Raz has put it, detached. And yet a detached ILS and an ILS made by a committed participant in a legal system seem to have the same meaning. Finlay and Plunkett’s quasi-expressivism explains why.
In addition to offering quasi-expressivism as their own account of ILSs, they also offer it as Hart’s view. Although I cannot pursue the details here, I think they are clearly right. Most notably, their reading can make sense of the fact that Hart was probably aware of the Frege-Geach problem before writing The Concept of Law6 –something that makes an expressivist reading of him awkward.7
- For a more nuanced description of Hart’s conception of external and internal legal statements, see Luis Duarte d’Almeida, The Grounds of Law, in The Legacy of Ronald Dworkin 165, 175-86 (W. Waluchow & S. Sciaraffa eds., 2016). Finlay and Plunkett’s understanding of the distinction is somewhat different from Hart’s. As they see it, an external legal statement speaks of a relationship to an identified rule of recognition of a legal system (such as “In China, it is the law that L.”) ILSs, by contrast, are unrelativized to a legal system (“It is the law that L.”) P. 59. I do not think their definition is adequate, for relativized statements can still be ILSs. I cannot pursue the matter here, however.
- Technically, Hart held a form of non-cognitivist expressivism concerning ILSs. See Kevin Toh, Hart’s Expressivism and His Benthamite Project, 11 Legal Theory 75, 78-81 (2005). It is not essential to expressivism that the relevant sentence expresses a non-cognitive state like an affect.
- See supra Toh note 2.
- Note that Hart’s view might have been that an ILS has expressive and descriptive meaning. See Pp. 49-86. But its use of normative language is tied to its expressive function.
- It may not be true or false as a matter of social fact, however. I would argue that the statement is about the relationship between abstract objects, rather than social facts. See Michael S. Green, The Semantic Thesis in Legal Positivism, in Cambridge Companion to Legal Positivism (Torben Spaak & Patricia Mindus eds., forthcoming). Finlay and Plunkett are somewhat equivocal on this matter.
- See supra Toh note 2, at 103-05.
- Thanks to Stephen Finlay for helpful comments.