To the dictionary ... and beyond!

Is there a way for a judge to "look up" the ordinary meaning of a word in a statute other than by using a dictionary?

This question underlies the provocative opinions in a case decided by the Utah Supreme Court on Friday. The dispute among the Utah justices may presage similar issues that courts elsewhere will confront.

The possibility of moving beyond the dictionary (or its online equivalents) comes from the academic language study known as "corpus linguistics." The term refers to the analysis of language through how it is actually used in a body of texts that come from real-world usage. In recent years, through computerization and the Internet, such collections of texts - the "corpus" or "body" of usage that is studied - have become far more extensive and accessible than they were a generation ago.

Utah v. Rasabout presented a standard issue of statutory interpretation for a court. The criminal defendant had fired 12 shots from a semi-automatic pistol at a rival gang member's house and car, leading to the shooter's conviction on twelve counts of unlawful discharge of a firearm.

The Utah Supreme Court had to determine whether the 12 counts should have been merged into one. This depended largely upon the meaning of the word "discharge" as used in a Utah statute: Was each shot a discharge, or was the entire series of shots that unloaded the gun a single discharge?

Using traditional statutory analysis, the majority held that each shot was its own crime. Under state law, the meaning of a statute depended in part upon the ordinary meaning of its words. The court determined the ordinary meaning of "discharge" by relying on the definitions in two standard dictionaries, other clues from the text of the statute, and the reasonableness of criminalizing each shot.

But the legal crossfire in the case came from a 40-page concurring opinion by Justice Thomas Lee, a former Brigham Young University law professor. Lee reached the same conclusion as the majority, yet he did so through corpus linguistics, along the way providing a full-throated defense of the approach.

Lee pointed out that dictionaries actually contained multiple definitions of discharge, and one ("to empty of a cargo: unload") supported the defense view. He argued the majority's choice between these definitions came from its "linguistic intuition" as to the ordinary use of the word.

This use of intuition, Lee wrote, was entirely appropriate. However, Lee believed that judges should transparently acknowledge that they were relying on such intuition and, then, "check [that] intuition against publicly available means for assessing the ordinary meaning of a statutory phrase." He did this in two ways.

First, Lee performed a "Google News" search and determined from the documents retrieved that when "discharge" is used in conjunction with a firearm, it is almost always used to refer to a single shot, rather than emptying all of the bullets in the magazine.

Even this step was unusual for judicial statutory analysis, though a prominent judge has done something similar. In United States v. Costello, 666 F.3d 1040 (7th Cir. 2012), Judge Richard Posner of the 7th U.S. Circuit Court of Appeals determined that the ordinary meaning of "harboring" a person connoted concealing him in some manner (not just giving him a place to stay) through comparing the relative number of "hits" that Google searches of various phrases using the word "harboring" revealed.

Lee criticized Posner's approach for relying too much on arbitrary search terms created by the judge. For instance, Posner searched for "harboring fugitives" but not "harboring aliens," which presumably affected the number of hits he received. Lee also pointed out that a Google hit search relies on results that are based on an unknown composition of underlying documents and an unknown search algorithm used by Google that cannot necessarily be routinely replicated by future users. Lee admitted that his Google News search (in which he analyzed the portions of the texts retrieved, rather than just the number of hits for search terms) suffered from similar, though reduced, flaws as those he saw in Posner's approach.

For that reason, Lee performed a second search, using the Corpus of Contemporary American English (COCA), which is curated by an accomplished linguistics professor at Brigham Young University. This corpus is arguably the most extensive such collection of English, containing over 410 million words from between 1990 and the present, "equally divided among spoken, fiction, popular magazines, newspapers, and academic texts." A COCA search is the type of analysis that many academic linguists would use to determine the ordinary meaning of a word as it is actually used in English discourse.

Lee pointed out that a COCA search has advantages over Google that are not limited to the known and balanced composition of the underlying corpus. The user need not use search terms, but can search for words within a certain distance of another (for instance, "discharge" within five words of "firearm"). The search is much like a Westlaw or LEXIS search, in that the searcher can examine the texts retrieved to see how the search terms are used. And in a COCA search, a judge or attorney can save and link to her searches, so others can easily replicate and examine them.

Lee discussed the results of his COCA analysis, concluding that the analysis "provides strong confirmation of the basis of our holding in this case. And it does so on the basis of a transparent database that is publicly available, created by linguists, and subject to replication by anyone seeking to confirm (or reject) my analysis."

The Rasabout majority attacked Lee's approach. One basis for criticism was case-specific: The majority claimed that because the parties had not briefed or argued about corpus linguistics, Lee should not have addressed it. Lee responded that corpus analysis, like independent legal research, was something that a court can perform on its own. Regardless of which view is correct, this criticism would be inapplicable in a case where such briefing was provided or invited.

Other criticisms were more enduring. The majority argued that judges are not experts linguists and thus may lack the ability to engage in corpus linguistics. Lee responded that judges are experts at statutory interpretation, and that judges in fact perform corpus analysis "intuitively in our minds" when determining ordinary meaning. The practice of using a language database is "not rocket science" and instead is "like math" - something everyone can do at a basic level, even if there are levels that only experts can perform. Lee was advocating the former: "using a calculator instead of doing it in our heads."

Additionally, the majority argued that corpus linguistics would present too great a burden on courts and litigants, perhaps necessitating "dueling linguistics experts" in every case. Lee called this argument a "straw man," noting that few cases actually need corpus linguistics; for example, where there was only one applicable definition in a traditional dictionary, there would be no need to delve further. When corpus linguistics is called for, Lee believed, it typically would not be difficult to present.

As well, the majority pointed to some concerns about Lee's particular analysis, including that he excluded several results because he could not tell which view they supported. The majority argued that this called into question the "statistical significance" of the sample in Lee's COCA analysis. Lee responded that the majority's approach was rooted merely in its own view of language use and "you can't get more insignificant than a data point of one - particularly if the data point is not unbiased." Lee claimed that his objective analysis of COCA language data nevertheless was better than "subjective linguistic intuition of an individual judge (or a panel of them)."

In Rasabout, Utah Chief Justice Matthew Durrant, authored a concurring opinion that seemed to mediate between Lee and the majority. Durrant rejected Lee's approach in the particular case, where the parties had not briefed corpus linguistics, but cautiously indicated a willingness to consider using the analysis in the future.

Durrant wrote: "Because so few judges and, perhaps, even fewer members of the bar are familiar with corpus linguistics, I believe it is simply too soon to know whether the benefits of using this new tool warrant the increased expense." He said he "look[s] forward to our continued debate in this area. Should we elect to embark down this path, we should tread slowly and cautiously."

It is quite possible that the justices' debate in the Beehive State soon will be echoed in other jurisdictions in difficult cases of statutory interpretation. Over time, more judges and lawyers may become aware of the availability of analysis of "ordinary meaning" based on corpus linguistics. Many judges may react to this endeavor with the wariness of Durrant, but in some "ordinary meaning" cases, the bench may one day have to decide whether it wishes to move beyond the dictionary.