Note: If you’re coming back to this post after having read it already, be sure to note the additional update I’ve added, which comes right after the discussion that dates the W2 definition back to 1934.
A new hashtag popped up last week, #GorsuchStyle, devoted to spoofing Justice Gorsuch’s writing style, or at least the style that is on display in the opening sentences of his dissenting opinion in Artis v. District of Columbia (pdf): “Chesterton reminds us not to clear away a fence just because we cannot see its point. Even if a fence doesn’t seem to have a reason, sometimes all that means is we need to look more carefully for the reason it was built in the first place.”
I mention that, not because this post will be about Gorsuch’s writing style (it won’t), but because I’m not above a little clickbait-y coattail-riding. It’s not gratuitous clickbait, mind you. This post is about Gorsuch’s dissent in Artis. And it remains to be seen whether it’s effective clickbait. Part of me thinks that if your title includes the word lexicographic, maybe you’re doing clickbait wrong. On the other hand, #AppellateTwitter likes dictionaries, so maybe not.
What this post will be about is Gorsuch’s choice of which dictionaries to cite in his Artis dissent. As the title suggests, it will be heavy on lexicography, but it will also touch on what that choice says about whether Gorsuch is a snoot, like his predecessor was, and if so whether that ought to play a role in his decision about what dictionaries to cite. I also plan on doing a separate post to talk about the contrasting approaches to word meaning that are on display in Artis, both of them resonating, though in opposite ways, with what I’ve written about that subject (link, link).
Update: The second post can be found here.
As you may have gathered know if you’re familiar with some of the things I’ve written about word meaning, I’m not a big fan of the central role that dictionaries play in the way lawyers, judges, and legal scholars deal with issues of word meaning. For this post, though, I’ll put that dissatisfaction aside, and will treat the use of dictionaries as perfectly appropriate interpretive tools.
THE ISSUE of which dictionary to use is a recurring one in the academic literature about judges’ use of dictionaries. It’s usually discussed under the rubric of “dictionary shopping”—the practice of looking for the definitions that are most supportive of the result you want to reach. That practice is of course the norm for lawyers arguing cases, but it’s problematic for judges or legal scholars, who aren’t supposed to start out with a preferred outcome and then reason backward to the arguments that can support it. However, there is at least one circumstance in which selectiveness about the dictionaries that one cites can be appropriate: when the purpose of citing the dictionary is merely to show that a particular reading of the word in question is possible. And in fact that was Gorsuch’s purpose in relying on the definitions he cited; he was arguing that the statute was ambiguous.
Nevertheless, Gorsuch’s dictionary choices in Artis are subject to some significant criticisms in terms of what might be called lexicographic relevance: the definitions he relies on don’t necessarily shed much light on the meaning of the statutory language that was in dispute. And in that respect, the dissent is not the first time that judges have gone lexicographically astray.
TO SET THE STAGE for looking at Gorsuch’s dissent, I’ll sketch out what the Artis case was about and what the majority concluded.
Federal district courts have jurisdiction over state-law claims only in certain kinds of cases. Such jurisdiction exists where (among other things) the plaintiffs aren’t citizens of the same state as the defendants. Jurisdiction over state-law claims also exists where such claims are asserted along with claims under federal law, to which the state-law are closely related. That kind of jurisdiction is known as “supplemental jurisdiction,” and it is what was involved in Artis. In a case where supplemental jurisdiction is the only basis for jurisdiction over the state-law claims, if the federal-law claims are dismissed, the state-law claims are usually dismissed along with them, and they may then be filed in state court. In order to protect plaintiffs against the risk of having the statute of limitations expire while the case is pending in federal court, the supplemental-jurisdiction statute provides that the “period of limitations for any [such] claim shall be tolled while the claim is pending and for a period of 30 days after it is dismissed unless State law provides for a longer tolling period.”
At issue in Artis was what the statute means in saying that the limitations period is “tolled.” Does tolling act as a “Pause” button that simply stops the running of the limitations clock while the federal case is pending and then for the 30 days after it is dismissed? Or does the clock continue to run, but with a 30-day grace period that starts if and when the federal case is dismissed? The majority adopted the stop-the-clock interpretation; Justice Ginsburg wrote the opinion. Gorsuch (joined by Kennedy, Thomas, and Alito) favored the grace-period reading.
In support of its position, the majority cited the 6th edition of Black’s Law Dictionary (which, although the opinion didn’t mention the fact, was published the same year that the statute was enacted):
To bar, defeat, or take away; thus, to toll the entry means to deny or take away the right of entry.
To suspend or stop temporarily as the statute of limitations is tolled during the defendant’s absence from the jurisdiction and during the plaintiffs minority.
The Court didn’t quote the full definition, but instead summarized it as saying that “‘toll,’ when paired with the grammatical object ‘statute of limitations,’ means ‘to suspend or stop temporarily.’” It described this definition as “captur[ing] the rule generally applied in federal courts.” And it relied on a variety of other factors, including the patterns of usage that are found in other federal tolling statutes. All in all, the majority concluded, the textual argument for the stop-the-clock interpretation “overwhelm[ed]” the argument for the grace-period reading.
THAT BRINGS US to Gorsuch’s dissent (which was joined by Kennedy, Thomas, and Alito). Gorsuch argued that the statute is ambiguous: “Where, as here, a law instructs us to ‘toll’ a limitations period, we know it may be telling us to do one of (at least) two different things.” One of the possibilities is to suspend the running of the limitations period; the other, he said, is to “allow the limitations clock to continue to run[,] but defeat the effect of the clock’s expiration for an additional specified period of time.” This argument relied entirely on three dictionary definitions—more specifically, two definitions from English-language dictionaries and one from a dictionary of Latin:
The dictionary informs that to “toll” means “[t]o take away, bar, defeat, [or] annul.” See 18 Oxford English Dictionary 204 (2d ed. 1989); Webster’s New International Dictionary 2662 (2d ed. 1957) (“[t]o take away; to vacate; to annul”); Oxford Latin Dictionary 1947 (1982) (“tollere,” the Latin origin, means to “remove” or “lift”).
Let’s discuss these definitions one at a time, starting with the Oxford Latin Dictionary. That definition (or more accurately, that translation) is irrelevant, because it’s a well-known fact that the United States Code isn’t written in Latin. Yes, the English verb toll, as used in the supplemental-jurisdiction statute, can be etymologically traced back to the Latin verb tollere. But that, too, is irrelevant: a word’s etymology doesn’t reliably indicate what it means now (link, link, pdf (pp. 1939-1940)). When the word nice first entered the English language 700+ years ago, for example, it meant “foolish, stupid, senseless.” The word silly can be traced back to the Old English word gesælig, which meant “happy, fortuitous, prosperous.” Over time, its meaning changed, “from ‘happy’ to ‘blessed’ to ‘pious,’ to ‘innocent’ (c. 1200), to ‘harmless,’ to ‘pitiable’ (late 13c.), ‘weak (c. 1300), to ‘feeble in mind, lacking in reason, foolish’ (1570s).” Gorsuch is hardly the first judge to inappropriately invoke etymology. For example, the majority in Muscarello v. United States committed the same blunder. But the etymological fallacy’s persistence in legal interpretation is disheartening.
The other dictionaries Gorsuch cites are English-language dictionaries, which is an improvement, but they present significant problems of their own. To begin with, consulting general-purpose dictionaries is arguably inappropriate in Artis given that the use of toll that is in question is a purely legal one. Indeed, the dictionary definitions quoted by the dissent were prefaced by the notation “Law” (in the case of the OED) and “O. Eng. Law” (in the case of Webster’s). In fairness to Gorsuch, however, his citation of general-purpose dictionaries is consistent with the majority’s invocation of the notion of ordinary meaning: “In determining the meaning of a statutory provision, we look first to its language, giving the words used their ordinary meaning.” Maybe they meant the ordinary meaning of the language when it appears in a legal context? Who knows?
In any case, even putting aside the question whether it was appropriate to rely on general-purpose dictionaries, the specific definitions that Gorsuch cited were badly out of date, and therefore support the dissent’s position weakly at best.
AS I’VE NOTED, the supplemental-jurisdiction statute was enacted in 1990. The OED definition that Gorsuch cited is from that dictionary’s second edition, which was published in 1989, and which therefore seems at first glance to be up to date. But upon taking a closer look, the picture looks much different.
One of the OED’s distinctive features is the quotations it provides to illustrate each sense of the word being defined. Those quotations come from real-world texts, in contrast to the practice of many dictionaries of using made-up examples, and in some ways they provide the real substance of the dictionary. Consider this statement by the linguist John Sinclair, who pioneered the modes of corpus analysis that are relevant to legal interpretation:
Those dictionaries that stand as milestones in our cultural history use real citations: Dr. Johnson’s Dictionary of 1755 and the Oxford English Dictionary begun by Murray in 1878. They understood that the dictionary is really just a commentary on the examples; the examples have a justification of their own.
[John Sinclair, The Dictionary of the Future (1987.]
I’ve gone into this detail because the most recent of the examples that go with the definition cited by Gorsuch is from 1818:The lack of any examples less than 200 years old suggested to me that maybe the second edition’s definition had been carried over from the first edition without change. Further evidence for that conclusion was provided by the current entry on OED Online, which is identical to the one above, and which has a note saying, “This entry has not yet been fully updated (first published 1912).”
And it turns out that my hunch was correct. I have verified that the entry quoted in Gorsuch’s dissent is the same as what appeared in the first edition. And after this post had been mostly written, I learned that that the second edition represents nothing more than a merger of the first edition’s content with that of the four volumes of supplements that had been published in the interim.
[Update. Jesse Sheidlower sends the following correction:
The second edition is primarily a merger of OED1 and OEDS, the four Burchfield supplements (1972-86). But it did add a number of new words (about 5,000), and some entries were revised. In most of the sort of examples that you’d be interested in, it’s overwhelmingly likely that OED2 text either comes from OED1 or OEDS, but it’s not absolutely guaranteed.
So the OED definition cited by the dissent was first published 78 years before the supplemental-jurisdiction statute was enacted, and may have been based on data that predated the supplemental-jurisdiction statute by more than 170 years. That hardly instills confidence that the definition accurately reflects how the legal sense of toll was used during the second half of the 20th century.
The Webster’s definition presents essentially the same issue, but in a less extreme form. The dissent cites the second edition of Webster’s New International Dictionary (W2), which was published in 1934. Although the dissent’s citation gives a date of 1957, that probably means only that the copy consulted by Justice Gorsuch (or, more likely, his clerk) was from a 1957 printing that was identical to the 1934 edition, except for the addition of a New Words section. I’m confident in saying that, for two reasons. I have gotten second-hand confirmation that the definition in the original 1934 printing was the same as the definition quoted by Gorsuch. And even before I received that confirmation, the facts of which I had personal knowledge made it nearly certain that the definition quoted by Gorsuch originated in the 1934 printing. That definition appears in my copy of W2, which bears a date of 1946 on the title page and a copyright date of 1945, and includes a New Words section for which the copyright date is “1939, 1945.” The fact that the new words were added as a separate section, rather than integrated with the original text, is evidence that the publisher did not redo the original typesetting (almost 3,000 large-format pages of tiny print).
[Update: Another correction from Jesse Sheidlower:
It’s not the case that 1945 W2 is identical to 1934 W2 + a new-words section (“is evidence that the publisher did not redo the original typesetting”). Usually, dictionary copyright updates involve a moderate number of (usually extremely minor) changes, so that perhaps dozens or a hundred pages will indeed be re-typeset. Most famously in the case of W2, the ghost entry for dord was removed in the 1941 printing, so that’s one big example of a difference between 1934 W2 and 1941-and-later W2s.
THUS, THE TWO DEFINITIONS cited by the dissent preceded the enactment of the supplemental-jurisdiction statute in 1990 by, respectively 78 years and 56 years. That fact alone has to make one wonder whether they accurately reflected how toll was used by the time 1990 came along.
And there are additional reasons for doubting the continuing adequacy of the dictionaries Gorsuch cites. First, the use of toll with regard to statutes of limitations earlier than 1934 may have been infrequent. It appears (from a Westlaw search) that before 1934, toll, tolls, toll, or tolling was used within 35 words of limitations in only eleven cases, of which only two were from before 1912. (There were five such cases in 1934 and one in 1912, but I assume that by those dates the definitions had already been finalized.) In contrast, when the same search is conducted for the period 1980 through 1989, it returns 642 cases
[Update: The previous paragraph turns out to be incorrect—the low number of hits that I was getting was due to the way I was doing the date-range filtering. When I fixed that problem, the search results didn’t give me reason to think there was a major change in the way toll was used after W2 was published. However, they do raise the question whether the W2 definition accurately reflected the range of early/mid-20th century usage. I’ll explain the basis for that statement in my second post about Artis, which I’m currently working on. ¶ My apologies for the mistake.]
Second, the dictionary definitions that were the most closely contemporaneous to the statute’s enactment differ from those cited by Gorsuch. And those differences suggest that, at least with respect to statutes of limitations, the definitions cited by Gorsuch were out of date [update: or not fully accurate]:
Black’s Law Dictionary 5th ed. (1979), 6th ed. (1990):
“To bar, defeat, or take away; thus, to toll the entry means to deny or take away the right of entry.
“To suspend or stop temporarily as the statute of limitations is tolled during the defendant’s absence from the jurisdiction and during the plaintiffs minority.”
Note: The definition above represents a change from the revised 4th edition (1968), in which the second paragraph read as follows: “To toll the statute of limitations means to show facts which remove its bar of the action.”
Random House Unabridged, 2nd edition (1987-1997; see note below):
“To suspend or interrupt (as a statute of limitations)”
Note: From what I can tell from my copy, the second edition was originally published in 1987 and then revised and updated in 1993, 1996, and 1997. I don’t know when this definition first appeared.
I will leave it for readers to work out for themselves the ways in which these definitions cast doubt on the continuing validity of the ones cited by the dissent.
But what about other dictionaries? With one exception, there is nothing. The relevant use of toll doesn’t appear in the New Oxford American Dictionary (2005), American Heritage Dictionary (4th ed. 2000, 2006), Merriam-Webster’s Ninth New Collegiate Dictionary (1984), or Webster’s New World Dictionary, Second College Edition (1982). While I don’t have copies of any editions of these dictionaries that came out closer to 1990, it seems unlikely that they would be any different. (I don’t think there was an earlier edition of NOAD.)
Of the major American dictionaries, that leaves only Merriam-Webster’s Third New International Dictionary (W3), which came out in 1961, and which did include the relevant sense of toll. Its definition was similar to the one in W2: “to take away : make null : REMOVE ‹⁓ the statute of limitations›.” This definition, being 27 years newer than the one in W2, was certainly more up-to-date than the latter. How much difference that makes is a judgment call.
NOW THAT THE DISSENT’S DEFINITIONS have been put into lexicographic-historical context, what are we to make of the decision to cite the OED and W2, which we now know aren’t really entitled to much weight?
With regard to the OED, the decision seems on its face to be justifiable. The citation was to the second edition, whose publication date was 1989, the year before the statute’s enactment. Moreover, the citation was to what we now call the hard copy of the second edition (although it sounds a little weird to use that phrase to refer to a 20-volume set of books). That’s relevant because it might not have been obvious from the physical book, with its tiny print, that the most recent example sentence was from 1818; that fact is much easier to see if you consult the online version, as I did. If the age of the most recent example been noticed, it might well have alerted Justice Gorsuch and his law clerks to the issue of up-to-dateness. Moreover, the online version contains an even bigger red flag that might not be as noticeable in the printed volume: the note saying, “first published 1912.”
So although I assume neither Justice Gorsuch nor anyone in his chambers realized that the OED definition was outdated, I’m prepared to rule that the Justice and his staff acted consistently with the standard of care that applies to justices of the Supreme Court, and by extension to all other U.S. judges. Unfortunately, that standard of care is lower than it should be. But that’s a topic for another day.
And that brings us, finally, to Justice Gorsuch’s decision to cite W2, from 1934, rather than W3, from 1961. What’s up with that? It has nothing to do with how the case was briefed; the only citation to either dictionary was in the respondent’s brief, which cited W3. And the two definitions are quite similar: “[t]o take away; to vacate; to annul” versus “to take away : make null : REMOVE ‹⁓ the statute of limitations›.” If anything, W3’s definition would provide more support than W2’s for the argument that the statute was ambiguous, because it specifically mentioned statutes of limitations. So why would he cite the older dictionary?
The only explanation I can think of is that Justice Gorsuch shares Justice Scalia’s irrational aversion to W3.
If you know about the controversy over W3, you can probably figure out what led Scalia to have such an aversion, and Gorsuch (I assume) to share it. But for those who are unfamiliar with the controversy, here is Geoff Nunberg to start filling you in:
At 2,700 pages and almost 14 pounds, Webster’s Third was a literally weighty work, the product of over 700 editor-years of effort, the publisher boasted. But it was widely denounced for what critics viewed as a lax admissions policy: it opened its columns to parvenus like “litterbug” and “wise up,” declined to condemn “ain’t,” and illustrated its definitions with quotations from down-market sources like Ethel Merman and Betty Grable. That was reason enough for The Times to charge that Merriam had “surrendered to the permissive school” and that the dictionary’s “say as you go” approach would surely accelerate the deterioration already apparent in the language. In The New Yorker, Dwight Macdonald wrote that the editors had “made a sop of the solid structure of English,” and in an Atlantic article called “Sabotage in Springfield,” Wilson Follett called the Third a “fighting document” that was “out to destroy . . . every obstinate vestige of linguistic punctilio, every surviving influence that makes for the upholding of standards.” (The dereliction that most appalled Follett was the Third’s refusal to reject “that darling of the advanced libertarians,” the use of “like” as a conjunction.)
Scalia was very much on the side of the condemnors. He expressed his disapproval in his opinion for the Court in MCI v. AT&T (more information available here and here), and during oral argument in Taniguchi v. Kan Pacific Saipan, Ltd.:
JUSTICE SCALIA: Webster’s Third, as I recall, is the dictionary that defines “imply” to mean “infer”—
MR. FRIED: It does, Your Honor.
JUSTICE SCALIA: —and “infer” to mean “imply.”
It’s not a very good dictionary.
Even if one shares the view that of W3’s entries were sometimes wildly permissive (which I think is at best an oversimplification) it is hard to justify preferring W2 even in cases, such as Artis, where the definition in question is unobjectionable and the statute in question was enacted after the publication of W3. In such situations, the more recent edition is always preferable, all else being equal. The only possible explanation I can think of for citing W2 rather than W3 is that doing so is a way of signaling that one holds a attitude toward usage that is seen by some as virtuous—that one is a linguistic “stalwart,” a stickler, a snoot. And while engaging in that kind of signaling may be fine in some contexts, it seems unwise to do so if the result is to weaken the legal analysis you are trying to advance.
Scalia’s (and apparently Gorsuch’s) objection to citing W3 is also misguided as a matter of interpretive theory. Statutory interpretation requires judges to determine how the statutory language is likely to be understood by ordinary speakers of English, and that in turn requires judges to be sensitive to how the words in the statute are actually used by such speakers. Even Bryan Garner acknowledges that word meanings arise from the way words are actually used (and that W3 is a good dictionary, albeit one that he thinks should be used cautiously because its usage notes are, he believes, inadequate). It is an undeniable fact that there is variation in how some words are used. Some people use infer to mean essentially the same thing as imply; for them, that is an appropriate use of the word, even though many other people see it as an error. Successful communication requires that interpreters take the speaker’s (or text’s) usage as they find it—even if they regard that usage as incorrect.
Indeed, Scalia and his co-author Bryan Garner seem to acknowledge this principle, at least to some extent, in Reading Law. Despite Garner’s position on the that-versus-which controversy (use that, not which, in restrictive relative clauses) the book says that this “grammatical convention” is “unfortunately a weak basis for deciding statutory meaning.” Why? Because “while grammarians have sought heroically to establish this as a firm rule, they have been unsuccessful.” (p. 142.) Therefore, one cannot assume that a which-relative is intended to be restrictive. (For more discussion, see this post, from which this paragraph has been adapted.) And when that principle is generalized to usage issues more generally, it undercuts any argument that W3 should not be cited because it is thought to be unduly permissive.
Note: Some quotations in this post have been cleaned up.