CORPUS LINGUISTICS: MISFIRE OR MORE AMMO FOR THE ORDINARY-MEANING CANON?

• Author: Ramer, John D.
• Position: NOTE

Scholars and judges have heralded corpus linguistics--the study of language through collections of spoken or written texts--as a novel tool for statutory interpretation that will help provide an answer in the occasionally ambiguous search for "ordinary meaning" using dictionaries. In the spring of 2016, the Michigan Supreme Court became the first to use corpus linguistics in a majority opinion. The dissent also used it, however, and the two opinions reached different conclusions. In the first true test for corpus linguistics, the answer seemed to be just as ambiguous as before.

This result calls into question the utility of corpus linguistics. If the Michigan Supreme Court could reach opposite conclusions about the ordinary meaning of a word when both opinions used the same database, can one really consider corpus linguistics an effective aid to statutory interpretation? This Note first argues that the Michigan Supreme Court's majority opinion conducted a flawed search of the database and that the dissent's contrary conclusion was correct.

In addition, this Note contends that transparency--the greatest strength of corpus linguistics--outweighs the risk that judges may fail to use a corpuslinguistics database accurately because that transparency permits advocates, other judges, and legal scholars to review the court's analysis, aiding appellate review. Lastly, this Note includes recommended steps to guide the use of a corpus-linguistics database in the search for ordinary meaning.

TABLE OF CONTENTS INTRODUCTION I. THE BACKGROUND OF CORPUS LINGUISTICS IN THE LAW A. The Legacy of Corpus Linguistics in Statutory Interpretation B. The Michigan Supreme Court's Use of the COCA 1. Factual Background 2. The Majority Opinion's Use of the COCA 3. The Dissent's Use of the COCA II. REVIEWING THE MICHIGAN SUPREME COURT'S COCA ANALYSIS A. Search Methods B. Contextual Review III. THE COCA AFTER HARRIS A. Criticisms of the COCA B. Responses to the Criticism 1. Experts 2. Sua Sponte Use C. The Value of Transparency D. Steps to Use the COCA CONCLUSION INTRODUCTION

The judiciary says what a statute means (1) in order "to give effect to the law" a legislature enacted. (2) To this end, courts "read the statute according to its text." (3) If the legislature did not define a particular word, courts generally look to the ordinary meaning of the word used (4) at the time the legislature adopted the statute. (5) This interpretive method--commonly called the "ordinary-meaning canon" (6)--assumes that a word's ordinary meaning most "accurately expresses the legislative purpose." (7)

This inquiry sounds simple enough: courts merely ask how a word is (or was) ordinarily understood. As one professor put it, the ordinary meaning is "that which an ordinary speaker of the English language--twin sibling to the common law's reasonable person--would draw from the statutory text." (8) But putting the theory into practice can be difficult.

In the last few decades, lawyers and judges have increasingly used dictionaries to determine a word's ordinary meaning. (9) For example, the ordinary-meaning canon was on display in Muscarello v. United States, (10) when the U.S. Supreme Court consolidated two criminal cases. (11) In one case, Frank Muscarello was arrested after driving to a drug deal with a handgun locked in his truck's glove compartment. (12) In the other case, which had similar facts, the defendants had stored their guns in the car's trunk. (13) In both lower court opinions, the courts of appeals found that the defendants had violated 18 U.S.C. [section] 924(c)(1) because they "'carrie[d]' the guns during and in relation to a drug trafficking offense." (14) The issue was whether the ordinary meaning of the phrase "carries a firearm" included storing handguns in a vehicle's locked compartment. (15)

The Supreme Court looked to a number of sources in its attempt to discern the phrase's ordinary meaning. First, the Court looked to the Oxford English Dictionary. (16) Then it consulted Webster's Third New International Dictionary, the Random House Dictionary of the English Language Unabridged, the Barnhart Dictionary of Etymology, the Oxford Dictionary of English Etymology, the King James Bible, Robinson Crusoe, and Moby Dick. (17) The Court also cited articles from the New York Times, the Boston Globe, the Colorado Springs Gazette Telegraph, and the Arkansas Gazette. (18) After reviewing all these sources (and others), the Court determined that the ordinary meaning of "carrying a firearm" included driving a car with a gun in it. (19) The dissent also conducted an extensive search for the word's ordinary meaning by consulting a legal dictionary, several translations of the Bible, poems, and scripts from the film The Magnificent Seven and the television shows M*A*S*H and Sesame Street. (20) The ordinary-meaning analysis of both opinions seemed somewhat haphazard.

To bring more statistical analysis to the ordinary-meaning canon, then-law student Stephen Mouritsen, (21) who held a master of arts in linguistics with an emphasis in corpus linguistics, (22) thought of a better way. He proposed using corpus linguistics (23)--the study of language through collections of spoken or written texts, called corpora. (24) These collections typically take the form of digitized databases that are accessible through an online interface. (25) A user can search the corpus-linguistics database for a particular word or phrase to study how that word or phrase has actually been used in the texts collected in the corpus. (26)

One corpus-linguistics database is the Corpus of Contemporary American English ("COCA"), which "is the largest freely-available corpus of English." (27) The COCA contains more than 520 million words and is equally divided among transcriptions of spoken language, (28) fiction publications, (29) popular magazines, (30) newspapers, (31) and academic texts. (32) Professor Mark Davies, a corpus linguist, developed the COCA through his academic work at Brigham Young University. (33) The COCA is a "tagged corpus," (34) which means that each word is labeled in reference to its particular part of speech--for example, a noun or a verb. (35) The COCA does not, however, function like a search engine. The user cannot simply ask what the word means. Instead, the database allows a user to review many instances of the use of a word or phrase in the database's collected texts. (36)

Mouritsen argued that the COCA's utility in determining ordinary meaning exceeds that of a dictionary (37) because dictionaries provide a range of permissible meanings only (38): "A dictionary, it is vital to observe, never says what meaning a word must bear in a particular context. Nor does it ever purport to say this." (39) Mouritsen described a hypothetical, seemingly modeled after Muscarello, where Judge A and Judge B disagree about the ordinary meaning of a word, and both judges find support in different dictionaries. (40) As Mouritsen explained, assuming no additional evidence, the dispute is intractable because it is based on the judges' differing intuitions about the word's ordinary meaning. (41) In contrast, if Judge A uses a corpus-based analysis, then Judge B "will have numerous 'overt' bases" to challenge Judge A's corpus analysis. (42) In the first scenario, the disagreement is a metaphysical debate; in the second scenario, the disagreement is an empirical debate. (43) This empiricism is the chief benefit of using a corpus-linguistics database as a supplement to a dictionary. The corpus is "essentially tell[ing] us what a language is like, and the main argument in favour of using a corpus is that it is a more reliable guide to language use than native speaker intuition." (44)

Since the publication of Mouritsen's note, (45) lawyers and judges have occasionally used corpus linguistics to aid in statutory interpretation. (46) Most recently, the Michigan Supreme Court became the first state supreme court to do so. (47) Interestingly, the dissent also used it but disagreed with the majority's corpus analysis. (48) Some judges are driving forward into the frontier with corpus linguistics. (49) Some judges think the databases are too complicated to use and should only be considered when the parties cite to a database in their brief. (50) Others think judicial recognition of corpus linguistics will force parties to hire linguistics experts, which will unduly add to the costs of litigation. (51)

This Note analyzes the judicial use of corpus linguistics in statutory interpretation through the lens of the Michigan Supreme Court's decision in People v. Harris. (52) Part I explains the evolution of using corpus linguistics--specifically the COCA--in statutory interpretation and examines the Harris opinions' corpus-linguistics analyses. Part II contends that neither opinion in Harris conducted a complete analysis of the COCA but that the dissent reached the correct result. And Part III recites some common criticism of the COCA, addresses this criticism, and briefly suggests practical steps for using the COCA.

1. THE BACKGROUND OF CORPUS LINGUISTICS IN THE LAW

   Part I traces the history of corpus linguistics in statutory interpretation. Section I.A discusses the emergence of corpus linguistics in scholarship, advocacy, and judicial decisionmaking. Section I.B examines the Michigan Supreme Court's use of corpus linguistics in People v. Harris by detailing the different types of searches that both the majority and the dissent used in their respective analyses.

   1. The Legacy of Corpus Linguistics in Statutory Interpretation

      Before the COCA appeared in any judicial opinions, it debuted in an amicus brief before the U.S. Supreme Court. In FCC v. AT&T Inc., (53) the Court had to determine whether corporations had "personal privacy" or if the phrase applied only to humans. (54) The Project on Government Oversight ("POGO") submitted an amicus brief that used multiple corpora to determine...