“The language of the law” is not actually a language

THE NATURE OF LEGAL LANGUAGE has been a recurring subject of discussion, within applied linguistics and (U.S.) legal academia. The latest contribution to that discussion is a recently-posted draft paper by John McGinnis and Michael Rappaport, titled The Constitution and the Language of the Law. (h/t Legal Theory Blog)

McGinnis and Rappaport are the primary advocates of an approach to constitutional interpretation known as original-methods originalism, under which courts today are to apply the interpretive methods that prevailed at the time of the framing (pdf). Their new paper argues that original-methods originalism is supported by the fact that (as they see it), the Constitution is written in “the language of the law.”

Although Larry Solum, of Legal Theory Blog, calls the paper “important and brilliant,” I’m afraid that I find its primary argument to be pretty seriously flawed. [UPDATE: McGinnis and Rappaport have responded to this post, and I have replied to their response.]

I’m going to talk here about two related aspects of the paper that I think are problematic. One is its treatment of “the language of the law” (a phrase that I will henceforth capitalize whenever I use it in the way that McGinnis and Rappaport do). McGinnis and Rappaport come close to treating The Language of the Law as a full-blown language on the order of French and Japanese, which I don’t think is justified by the facts. The other major problem that I see lies in the analogy that the paper draws between the rules of legal interpretation and what it calls the “interpretive rules” of ordinary language (which are better described as the cognitive processes involved in the comprehension of utterances and texts). This analogy, which plays a key role in McGinnis and Rappaport’s argument, is invalid because each of the things that they are analogizing is fundamentally dissimilar from the other.

BUT BEFORE I TALK ABOUT The Language of the Law, I have to say a few things about the word language. As is common among high-frequency words, language is polysemous—i.e., it has multiple different senses that are related to each other. For example:

• Language can be used to refer to the system of communication that humans (but no other species) are capable of, in which small units of meaning called words are combined into larger units of meaning called phrases, clauses, and so forth. We can talk about language as a cognitive capacity, about the evolution of language, and about children acquiring language as infants. Let’s call this the “system-of-communication” sense of language. When I use language in this sense, I’ll sometimes include a disambiguating subscript, like this: language_system.

• The noun phrase a language can be used to denote the particular type of language_system that used by a given population group; for example, English, Chinese, or Urdu. I’ll refer to this as the “armed-dialect” sense of language (language_dialect), based on the fact that because there is no principled distinction between a language and a dialect, linguists sometimes say that a language is a dialect with an army and a navy. Two points to note about this sense: First, any use of a language_dialect is of necessity a use of language_system, and (mostly) vice versa. Second, of the four senses of language that I’ll discuss here, this is the only one in which the word typically appears as a countable noun.

• Language can be used as a way to refer to particular uses (or categories of uses) of language_{system/dialect}, especially when one wants to focus on the characteristics of those uses. Thus, we can describe “the language of [a particular text]” as being elegant, or sloppy, or verbose. A literature professor can talk about “Shakespeare’s language,” meaning the ways in which Shakespeare deployed Early Modern English. And of course we often use the phrase legal language to refer to the ways in which lawyers and lawmakers deploy Modern English. I’ll refer to this as the “characteristics-of-use” sense (language_{use-characteristics}). Note that this sense of the word can be used both to refer to a kind of language_{use-characteristics}, as in Legal language is difficult for nonlawyers to understand, or to a particular instance of such language_{use-characteristics}, as in That document contains a lot of legal language. And note also that any use of language_{use-characteristics}, will of necessity amount to a use of a language_dialect, and therefore of language_system.

• Language can be used to refer to a particular chunk of text, as when we talk about the statutory language that is at issue in a case. I’ll call this the “chunk-of-text” sense (language_chunk).

HAVING LAID THAT GROUNDWORK, let’s talk about The Language of the Law—both the phrase and the thing that the phrase refers to. My impression from reading the literature is that the phrase the language of the law is most often used in such a way that it is essentially a synonym for legal language, and means the language_{use-characteristics} of the law. And that is appropriate, considering that whatever label you apply to legal language, it is not a full-blown language_dialect in the same sense as Arabic and Hindi. As the late Peter Tiersma pointed out, “If we isolate what is distinctive in legal English, leaving out features of ordinary speech, what remains is far too incomplete to function as a language.” And of course nobody grows up speaking legal language as their mother tongue.

Occasionally, though, the phrase the language of the law is used as a way to talk about legal language as if it were a language_dialect:

“Lawyers and judges, when they are acting in their roles within a legal system, speak the language of the law.” (Link)

“[All] lawyers are at least bilingual. That is, they speak both their native language and the language of the law…. [However] we may be immersed in the language of the law, it is at best a second language for us.” (Link)

I don’t read these uses of the language of the law  as literally equating legal language with true languages_dialect. Rather, they are rhetorical tropes, and as such they don’t fit easily into any of the senses I described above. In fact, they show why one shouldn’t assume that the different senses of a word are separated by clear boundaries or that a particular use of a word can evoke only a single sense. (See pages 27-39 of the linked paper.) The uses of the language of the law in the quotes above can be seen as combining a weakened version of language_dialect with aspects of language_{use-characteristics}.

So what do McGinnis and Rappaport mean when they use the phrase the language of the law—do they regard The Language of the Law as a language_dialect? Although they seem to come down on both sides of the question, they ultimately take the idea of it being a language_dialect fairly seriously. On the one hand, they say that The Language of the Law is “not wholly independent of ordinary language” (which is an understatement) and describe it as “an overlay on ordinary language.” But on the other hand, they make statements like these:

“It is clear that a language of the law exists—one that is employed by lawyers and others who are learned in the law.”

“A significant part of learning the law involves learning to speak, write, and interpret texts like a lawyer. These tasks involve learning a distinctive legal language.”

“The language of law is but one of many technical languages.”

“The case for [the existence of] a language of the law can be made under either a broad or narrow understanding of what is language.”

“What comprises the language of the law? As with other technical languages, the language of the law can be thought of as a language that builds on ordinary language, but then substantially supplements and modifies it.”

These uses of the language of the law strike me as more literal than those that I quoted previously. And that impression is confirmed by the substance of McGinnis and Rappaport’s argument, which depends on analogizing The Language of the Law to ordinary language—an issue that I’ll turn to now.

McGINNIS AND RAPPAPORT OFFER THEIR ANALOGY (and their overall argument about The Language of the Law) in support of the theory of constitutional interpretation they advocate: original-methods originalism. They argue that the Constitution is written in The Language of the Law and that as a result it should be interpreted using the interpretive rules that prevailed when it was adopted. The analogy between The Language of the Law and ordinary language represents the crux of the strong version of their thesis, which relies on what they call a “broad conception” of language in general and of The Language of the Law in particular. (The weak version relies, naturally enough, on a “narrow conception” of those things.)

The analogy focuses on the supposed parallel between the rules of legal interpretation and what McGinnis and Rappaport call the “interpretive rules” of ordinary English. The argument goes essentially as follows:

1. The components of a nontechnical language (i.e., of ordinary language) include word meanings and “rules that [language users] use to understand speech.” Those rules “include not only grammatical rules but any rules that tell speakers how to interpret the language.” Thus, “language should include all the background rules that help decode the communication of a speaker or writer.”

2. The Language of the Law has rules of its own, including “the legal interpretive rules which tell the speakers how to interpret the language.” Legal interpretive rules are like the interpretive rules of ordinary language, in that each set of rules is part of the “regularities” of that language. And just as the interpretive rules of ordinary language are part of that language, the rules of legal interpretation as are part of The Language of the Law.

3. The Constitution is written in The Language of the Law as it existed in 1789. Because the rules of legal interpretation as they existed then were part of the Language of the Law, they are rules of the language in which the Constitution is written. And just as the interpretation of an ordinary-language text is governed by the rules of that language, including its interpretive rules, the interpretation of the Constitution is governed by the rules of The Language of the Law as of 1789, including its interpretive rules.

To understand this analogy, it’s necessary to understand what McGinnis and Rappaport mean by what they call the “interpretive rules” of ordinary language. Unfortunately, their attempts at explanation don’t tell us anything useful: interpretive rules are rules (other than the rules of grammar) “that tell [language users] how to interpret the language”; they are “the background rules that help decode the communication of a speaker or writer.” These are essentially tautologies: interpretive rules are rules that contribute to the process of interpretation.

But we can get some idea of what McGinnis and Rappaport mean by looking at their paper’s examples of ordinary-language interpretive rules. They offer two examples:

“[It] is normally thought that a speaker or writer will not contradict himself in a document. Thus, interpreters try to understand different provisions as consistent.”

“An example of an ordinary language interpretive rule is the rule that one assumes that an ambiguous term should be construed in accordance with the subject of an utterance.” The footnote to this sentence provides an illustration of this assumption in action: “[If] a speaker is talking baseball, then the statement ‘that should work on the diamond’ should be interpreted to mean the baseball diamond rather than a gem.”

What these examples suggest is that what McGinnis and Rappaport call the “interpretive rules” of ordinary language include the cognitive process that are at work in connection with discourse coherence—i.e., that contribute to the sense that an utterance or text hangs together as a whole. Those processes often result in the comprehender mentally constructing a meaning that is richer than would be justified by the utterance or text alone. And both of McGinnis and Rappaport’s examples involve such processes.

It is unclear what else, beside discourse coherence, is encompassed in  the “interpretive rules” of ordinary language. However, pending clarification by the authors, it seems reasonable to assume that these “rules” include all the mental processes by which a comprehender’s understanding of an utterance or a text goes beyond the meaning derived solely from compositional semantics. Under that assumption, the “interpretive rules” would include pragmatic enrichments, implicatures, and inferences. In fact, those phenomena could well be considered to be aspects of discourse coherence, broadly considered.

Given this understanding of what the rules of legal interpretation are being analogized to, it seems hard to avoid the conclusion that they regard The Language of the Law as coming quite close to being a language_dialect. And even if that is not what they subjectively believe, that conclusion is essential to the validity of their analogy between legal interpretive rules and the ordinary-language “interpretive rules.”

As my earlier discussion of the issue suggests, I don’t see any basis for treating The Language of the Law as approaching the status of a language_dialect. While there are certainly ways in which legal language_{use-characteristics} is distinctive, the features that are relevant to legal interpretation relate almost entirely to word meaning: specialized legal terms and ordinary words that have specialized legal meaning. But a language is more than just a bag of words (or, if you want an image that’s more dignified, more than just a dictionary). If someone wants to argue that The Language of the Law is analogous to a language_dialect, they need to identify the linguistic features of The Language of the Law that make the analogy valid. And that is something that McGinnis and Rappaport make no effort to do.

THIS LEAVES THE QUESTION whether, even apart from whether The Language of the Law is (analogous to) a language_dialect, the rules of legal interpretation can be validly analogized to the “rules of interpretation.” The answer is no. The two types of “interpretive rules” differ fundamentally from one another, and to analogize the rules of legal interpretation to the “rules of interpretation” of ordinary language is a serious category error.

The cognitive processes involved in comprehending an utterance or a text—which is to say, the processes that McGinnis and Rappaport call the “interpretive rules” of ordinary language—occur auto­matically, effortlessly, and, for all practical purposes, instantaneously. Moreover, these processes occur below the level of conscious awareness and inaccessible to introspection. Legal interpretation, on the other hand, is a deliberative process by which the interpreter consciously thinks about how the utter­ance or text should be understood. This process takes time and requires cognitive effort. And although the process is obviously subject to cognitive biases, it is a process of explicit reasoning.

Considering these differences between the two types of “interpretation,” I don’t see any justification for analogizing them. Maybe there’s some underlying commonality that I’m not seeing, but if McGinnis and Rappaport think that such a commonality exists, they need to explain what it is and why they think it exists. Otherwise, it is unclear whether the strong version of their original-methods thesis (the one based on the broad conception of language) can survive.

THERE IS ALSO ANOTHER WAY in which the differences between the two kinds of “interpretation” are relevant here. The differences highlight the fact that the words interpret, interpretive, and interpretation, like the word language, are polysemous. They can be used to refer both to the conscious process of deliberation that underlies legal interpretation and to the automatic and effortless cognitive processes that underlie the comprehension of utterances and texts. Although it’s not unusual to use the word interpretation to refer to both processes, the effect in this context is to obscure the fact that the processes differ. Thus, I prefer to use interpret and its derivatives only with respect to legal interpretation, and to use the words comprehend and comprehension to refer to the cognitive processes by which utterances and texts are understood.

An issue of polysemy also exists with respect to the other half of the phrase interpretive rules. McGinnis and Rappaport use the word rule to refer to two different phenomena. The “rules” that govern the process of comprehension are in reality just generalizations about how the process usually functions. Thus, they are better described using a word such as “regularities,” which McGinnis and Rappaport do use several times. But whatever the label, the fact remains that these phenomena do not depend on obedience to explicit commands issued by some authoritative source. Rather, they are emergent properties that arise from the normal functioning of human cognition.

Legal interpretive rules, on the other hand, are very different. They are promulgated explicitly by actors vested with institutional authority, and they are owed obedience, or at least deference,  by other institutional actors. While the processes underlying comprehension are an inherent and essential part of using and understanding language, the rules of legal interpretation don’t come into play until the comprehension process is complete; their very existence depends on the interpreter’s ability to use and understand language.

SOMETIMES IT DOESN’T MUCH MATTER how we label things. A rose by any other name, and all that. But in some cases, labels do matter, because their use can lead to misdirection. The paper that I’ve been discussing is a case in point. While I don’t mean to suggest that McGinnis and Rappaport had any inappropriate intent, the fact is that the labels they use make it harder to see what their argument actually entails.