The Judicious Use of Corpus Linguistics
David Weisberg
The abstract of an article by Profs. Clark Cunningham and Ute Römer—“Can a President Be Impeached for Non-Criminal Conduct? New Linguistic Analysis Says Yes”—has been posted by Prof. Michael Ramsey, who agrees with the affirmative conclusion in the title. I too agree with that conclusion (see, here). But what I find most significant about the Cunningham & Römer article is that it is an instance of an extremely rare phenomenon: the appropriate use of Corpus Linguistics. I believe CL is almost always invoked inappropriately. Here is why I think the article is an exception to the rule.
In a paper published on SSRN, I posited what I call the Paradox of Originalism:
If the antiquity of the Constitution justifies the rebuttable presumption that some or all of the words or phrases in the Constitution have time-dated original meanings that differ from their current meanings, then the roughly equivalent antiquity of secondary literary materials that are roughly contemporaneous with the Constitution justifies the rebuttable presumption that some or all of the words or phrases in those secondary literary materials have time-dated original meanings that differ from their current meanings.
The Paradox, I submit, is self-evidently true. A few additional moments of reflection reveal that any originalist who presumes that, because of the antiquity of the Constitution, all the words and phrases in it might have time-dated meanings that differ from their current meanings, will inevitably be caught in an infinite regress.
The paradigmatic instance of the Paradox is found in Justice Scalia’s majority opinion in D.C. v. Heller, where he investigated the meaning in the Second Amendment of the words “Arms” (554 US at 581), “keep” (554 US at 582), and “bear” (554 US at 584). Justice Scalia began his examination of each of those words by turning to the 1773 edition of Dr. Johnson’s dictionary; he learned, e.g., that “Arms” is defined therein as “[w]eapons of offense, or armor of defense.” But that dictionary pre-dates the Constitution itself. Therefore, any conscientious originalist who entertains the presumption referred to in the Paradox (and why would Justice Scalia immediately turn to a dictionary published in 1773 unless he were entertaining that presumption?) would again have to resort to late-18th-century literary materials to determine the meaning of the words “weapons,” “offense,” “armor,” and “defense” in 1773. And, once those four words have been defined, the words in their own definitions will have to be defined, etc., ad infinitum.
It’s similarly self-evident that, if one presumes that all the words and phrases in the Constitution might have time-dated meanings different from their current meanings, the CL methodology also exemplifies the Paradox of Originalism and the infinite regress it generates. This is entirely predictable, because CL creates what are in effect custom-made dictionaries that are contemporaneous with an antique text.
Practitioners of CL believe that, when studying an antique text, the proper way to ascertain its meaning is to compile a corpus of contemporary literary materials in which key words or phrases in the target text appear, and then to determine, based on that corpus, what are the predominant or ordinary senses of the target text. But the literary materials in the corpus are by design roughly as ancient as the target text, and so the question arises: how can we be sure that we correctly understand the key words and phrases in the corpus? To achieve certainty, we must create new corpora corresponding to those new target terms, and now we’ve entered into an infinite regress.
There are, nevertheless, instances where one might very usefully consult contemporaneous literary materials, including dictionaries and corpora, to ascertain time-dated meanings of antique texts. Those instances, however, are not generated by presumptions about the mere possibility of time-dated meanings that differ from current meanings. Rather, they arise out of two specific circumstances of which I am aware (and there may be other circumstances of which I’m not aware).
One circumstance would be where a word or phrase is so archaic and obsolete that we have no clear idea what it meant when it was used in the Constitution. For example, Article I, Section 10 discusses “Letters of Marque and Reprisal” and “Bill of Attainder,” among other things. These phrases are obviously legal terms of art that were current in 1789, but are not current today. It makes perfect sense to flesh out their meaning by resorting to contemporaneous literary materials, but not because we entertain some general presumption about all the words and phrases in the Constitution. It is, rather, because we are very much aware that we have no clear understanding of what those particular phrases, which are now obsolete, meant in 1789.
A second circumstance would arise where the language of the Constitution, taken in full context, seemingly contradicts our contemporary understanding of its meaning. An example would be the language of Article II, Section 4, which recites that impeachable offenses are: “Treason, Bribery, or other high Crimes and Misdemeanors.” This is, of course, the phrase that Cunningham & Römer examine in their article. And it certainly invites close examination, because in today’s legal terminology a “misdemeanor” is a relatively minor offense, and a “high misdemeanor,” to which the constitutional text seems to refer, would be a present-day oxymoron.
In sum, both Justice Scalia’s rather old-fashioned reliance on literary materials such as Dr. Johnson’s dictionary and the new-fangled methodology of CL will inevitably fall afoul of the Paradox of Originalism, if those methodologies are deployed because someone entertains a presumption about the mere theoretical possibility of time-dated meanings that differ from current meanings. But both those methodologies might properly be used either when we reasonably believe we have little or no understanding of long-obsolete words and phrases in the Constitution, or when our understanding of the words and phrases, considered in context in the Constitution, makes no sense.