« Mark Pulliam on Sunstein & Vermeule on the Administrative State
Michael Ramsey
| Main | Ashlee Paxton-Turner: Continental Originalism
Michael Ramsey »

12/27/2020

And, on the Third Day, God Created Dry Land and Refuted Corpus Linguistics
David Weisberg

Professors James Cleith Phillips and John Yoo have published a paper arguing that corpus linguistics generates a negative answer to the question: Does the Antiquities Act of 1906, which authorizes the president to declare as national monuments objects of historic or scientific interest situated upon lands owned or controlled by the federal government, authorize the president to make such declarations with regard to objects on the ocean floor?  With all due respect, their paper vividly illustrates why corpus linguistics is dangerously unreliable as a method of legal analysis.

First, accepting the criterion that Phillips and Yoo themselves adopt, corpus linguistics is unsatisfactory on its face.  This is how they summarize the methodology:

[I]f 90% of people would understand a word to mean X in a particular context, and 10% would understand a word to mean Y in that context, textualism would declare X to be the ordinary meaning, as long as X and Y did not overlap or one was not subsumed within the other. Since we cannot survey Americans from the 1900s (and surveying people has its own problems for determining meaning), we will need to turn to corpus data to see if there was a dominant sense of the relevant terms from the Act, at least dominant in how frequently it was used. If there is, that dominant sense is a particular term’s ordinary meaning.

The notion that dominance “in how frequently [a term] was used” in a certain context can somehow be equated with “dominant sense” or “ordinary meaning” is false on its face.  Frequency does not determine or even alter meaning.  Consider the following. 

It’s certain that, when President Trump was impeached and on trial, a survey of written materials published in the U.S. would have found that the word “misdemeanors” overwhelming appeared in the context of “high crimes and misdemeanors,” in which context it meant serious, substantial wrongdoing.  Nevertheless, we all understood that that word also retained what had formerly been its more common usage: a relatively minor, insubstantial infraction.  The frequency of one usage would not have obliterated or changed the meaning of the less-frequent usage.  Even during Trump’s Senate trial, crossing against the light was still a misdemeanor.

Of course, in 1906 (and, I would bet, today) the word “land” or “lands” was used infrequently in discussions of the seafloor.  But it is one thing for legislators not to give any substantial thought to the seafloor, and it is a very different thing for them deliberately to exclude the seafloor from the scope of legislation.  In fact, if they did deliberately exclude the seafloor, one would expect that there would be more frequent use of “land” in that context than if they had never given the matter serious consideration either way.

This brings us directly to the question a statistical methodology, such as corpus linguistics, can never answer: If that’s what the legislators meant, why isn’t that what they enacted?  That is, why didn’t the 1906 Act provide that the president could declare monuments on “dry lands” of the federal government?  Is anyone going to argue that, in 1906, the phrase “dry lands” was not in common use?  The King James version of the Bible, generally accepted as the most widely published English text of all time, says that, on the third day of creation, God gathered the waters together and let “the dry land” appear.  Were legislators in 1906 ignorant of the Bible?  Moreover, the Oxford English Dictionary has a definition of the phrase “dry land”—“Land not submerged or under water; land as opposed to sea.”—which has been valid since the year 1225. This definition establishes that, over the last millennium, competent English speakers have at least implicitly understood that "land" refers to stuff that can be either submerged under water or not.  Whether above or below water, it's still "land".

There is absolutely no reason to believe that Congress intended to limit the Antiquities Act to objects on dry lands and to exclude objects on submerged lands; they could easily have used the phrase “dry lands” if they so intended.  Moreover, there is no reason to believe that underwater objects have less historic or scientific interest than objects on dry lands, even if legislators did not discuss that possibility.  In sum, corpus linguistics provides a clearly incorrect interpretation of the Antiquities Act.