Introductory note: In advance of the colloquium on corpus linguistics and the Second Amendment, all the participants submitted drafts of their blog posts, which were in turn distributed to the other participants. Those drafts raised a host of interesting issues—too many to be adequately addressed in the time allotted for the colloquium. I therefore prepared and circulated written comments on everyone’s draft ahead of time. I’ve updated and revised some of those comments in light of the discussion at the colloquium and of changes to some of the posts made after the colloquium. For anyone who is interested, these comments are available here.
Discussion of the corpus-based challenge to Heller has as far as I know focused almost exclusively on only one aspect of what the corpus data shows (Will Baude’s contribution to this series of posts is an exception), and as a result has understated the extent to which the evidence undermines Heller. What I want to do here is to broaden the discussion by pointing out important issues that have so far been neglected.
The discussion to date (apart from Will’s post) has concerned the meaning of bear arms, and most people have described the corpus data in terms that seem to accept the premise that bear arms was understood to mean ‘carry weapons,’ but with the qualification that the phrase denoted carrying weapons in the military. As I’ll discuss, though, that premise is inaccurate. The evidence suggests that bear arms was used almost exclusively in a variety of idiomatic senses having to do with the war and the military, not the literal carrying of weapons—precisely the opposite of what Heller concluded.
Discussion of the challenge to Heller has also neglected the light that the corpus evidence has shed on other parts of the Second Amendment, including especially the right of the people. The evidence for that phrase goes a long way toward showing that Heller was wrong in regarding the Second Amendment as protecting an individual right. And the evidence for well regulated militia, which I’ve recently looked at, similarly supports that conclusion.
I’ll talk about these issues in what follows, but given the limited space available I’ll only be able to scratch the surface. For the details, download the 2019 compilation of my original analysis, as well as my amicus brief opposing cert. in Folajtar v. Garland, which deals with well regulated militia.
The Supreme Court in Heller rejected the argument that bear arms was ordinarily used as an idiom conveying a military-related meaning, and instead based its interpretation on the separate meanings of bear and arms—or, more precisely, on dictionary definitions of those words. The Court was wrong to reject the idiomatic reading, because idiomatic uses accounted for 95% the instances in the corpus of bear arms and its grammatical variants (bearing arms, bore arms, etc.).
It’s nevertheless worthwhile to look at the data for bear and arms individually, because it shows that the picture painted by the definitions the Court relied on was oversimplified to the point of being misleading.
Based on those definitions, the Court interpreted bear as being essentially a synonym for carry. That view was widespread then time, and remains so today, but the corpus data paints a different picture. Although bear was sometimes used to denote the kind of carrying that the Court had in mind, such uses represented the exception rather than the rule. And the other ways in which bear was used differed significantly from how carry was used.
These differences may well reflect the fact that, as explained in a 2014 book by the principal etymologist for the Oxford English Dictionary, carry had by the end of the 1600s largely replaced bear as the verb generally used to convey the meaning that Heller attributed to bear. And if that’s correct, the view of bear as being synonymous with carry may have arisen from American usage after the Second Amendment’s ratification. So in seeking to discover 18th-century meanings more than 200 years after the fact, the fact that we think we understand how a word was used doesn’t necessarily reflect the reality.
And that warning applies as well to arms, albeit to a lesser extent. While it’s true that arms was often used as meaning ‘weapons,’ it was also used in a host of military-related figurative senses. That’s important because, in conjunction with the evidence regarding bear, it provides a reason to believe (even without looking at the relevant corpus data) that Heller was mistaken about bear arms.
And such a belief would be borne out by the data. I analyzed 535 deduplicated concordance lines from the period 1760-1799 (see this post for a discussion of the search criteria and the scope of my analysis). Out of that total, I categorized only 11 as unambiguously using bear arms in the nonmilitary sense that Heller regarded as its “natural” meaning. (In contrast, that sense accounted for more than half the uses of carry arms.) Dennis Baron’s results were similar, and based on what Josh Blackman and James Phillips have said to date (link; link), their results seem to have been generally consistent with Dennis’s and mine.
As of this writing, not a single gun-rights advocate has disputed either my reading of the data or Dennis’s. In fact, there has to my knowledge been only one article, blog post, or tweet that even partially disagrees with either Dennis or me: The “Weaponization” of Corpus Linguistics: Testing Heller’s Linguistic Claims (2020), by Josh Jones, who was clerking on the Utah Supreme Court when the article was published. Although I don’t regard Jones as a gun-rights advocate, my impression based on reading between the lines is that he is generally sympathetic to the analysis and result in Heller. I point that out because despite what I suspect are his priors, his reading of the data is broadly consistent with mine. While Jones categorized more uses of bear arms as nonmilitary than I did, he nevertheless concluded that idiomatic military uses were roughly three times as frequent as those that he regarded as literal. And that conclusion is at odds with Heller.
Even assuming Jones was correct in his judgments categorizing uses as literal and nonmilitary, his conclusions would be insufficient to defend Heller against the corpus-based challenge. With idiomatic military uses outnumbering literal/nonmilitary uses even by his count, bear arms would at best (from the viewpoint of gun-rights advocates) be ambiguous. And under the analytical framework followed by Heller, consideration of the prefatory clause would be appropriate in order to resolve the ambiguity. At that point, it seems to me, Heller’s interpretation is no longer viable.
Although Heller saw the prefatory clause as being consistent with its reading of the operative clause, that conclusion is undermined by the Court’s misunderstanding of 18th-century usage. Moreover, there is historical support for reading the prefatory clause as being consistent with the view of the operative clause that I think the corpus evidence supports. (For example, see the work of Patrick Charles and Saul Cornell.) Indeed, that reading strikes me as being more consistent with the prefatory clause than Heller’s.
the right of the people
In interpreting the right protected by the Second Amendment as one that belongs separately to each of the persons who together comprise “the people,” the Court in Heller adopted the kind of reading that is referred to in linguistics as “distributive” (as opposed to “collective”). The difference between the two types of readings is illustrated by comparing the following examples:
(1) The workers picked up the boxes (will generally be understood distributively: each worker individually picked up at least one box).
(2) The workers picked up the piano, (will generally be understood collectively: all the workers together picked up the piano).
The Court based its distributive interpretation on the way that the right of the people was used elsewhere in the Constitution, and while there are reasons to criticize that part of the opinion, I want to talk here something that the Court didn’t consider: the way the right of the people was used in writings other than the Constitution.
In an effort to answer the question that Heller didn’t ask, I reviewed the relevant data for the period 1760 through 1799. I found that there were more than 11 times as many uses in which the right in question was unambiguously collective (69 uses) as those in which it was unambiguously individual (6 uses). My criterion for categorizing a use as collective was whether the right was one whose exercise required the collective action of multiple people. For example, the right of the people—
to frame their own Constitution and form of Government
to alter the line of succession to the throne
to amend, and alter, or annul their Constitution, and frame a new one
to be represented in the legislature
to call a Convention at any time
to change their government & give it the form they please
The results were roughly similar for the phrases the people have a right and the people’s right: I categorized 63 as unambiguously collective and 17 as unambiguously individual.
This is evidence that formulations such as the right of the people to [X] was used with respect to collective rights more often than with respect to individual rights, and that those rights were fundamental political rights. And that’s important because it makes it significantly more likely that the Second Amendment would have been understood as protecting a right that was collective and political than one that was individual and apolitical.
well regulated militia
My original analysis, which I completed in 2019, did not deal with well regulated militia. However, I addressed that issue earlier this year, in an amicus brief opposing cert. in Folajtar v. Garland. And as discussed in the brief, there is a good deal of evidence pointing to the conclusion that the concept of “regulating” the militia was understood as a general matter to entail governmental regulation, and that in the case of “well regulated militias” in particular, the regulation was similarly understood as being imposed by the government.
Much of the relevant evidence comes from colonial and state militia statutes (see the Source Note, below), which were often titled as Acts “for regulating the militia,” “for the regulation of the militia,” etc., and which often identified the statute’s purpose as being to ensure the maintenance of a “well regulated militia.” Such statutory language reflects an understanding that it was colonial and state governments that performed the function of regulating the militia and that it was such regulation that was the hallmark of a “well regulated militia.”
Additional evidence comes from other sources, perhaps the most important of which is the Articles of Confederation, which required every state to “always keep up a well regulated and disciplined militia, sufficiently armed and accounted.” That may have been an especially salient part of the historical background against which the framing and ratification of the Constitution and Bill of Rights played out.
The understanding that a well regulated militia constituted a state-regulated militia was also reflected in statements by several of the Founders. For example, early in the Revolutionary War, George Washington complained to the governor of New Jersey about the lack of discipline in the New Jersey militia and urged that this problem be addressed by the enactment of “a well regulated Militia Law.” And in an address during his second term as president, he said that “the devising and establishing of a well-regulated militia, would be a genuine source of legislative honour” (cleaned up).
Statements reflecting a similar view of what a well regulated militia amounted to were also made by Alexander Hamilton, James Mason, Oliver Ellsworth, James Madison, and others. For details, see my brief in Folajtar.
I should note that I’m not the first person to have pointed out the evidence I’ve been discussing. But I think the evidence takes on added force when it’s considered in light of the corpus data—which brings me to the next issue I want to discuss.
The whole is greater than the sum of the parts
What the heading above refers to is the way in which everything I’ve been discussing hangs together as a coherent whole. The fact that bear arms was predominantly used in an idiomatic military sense is consistent with the evidence as to bear and arms. And that pattern of use is consistent with the evidence as to the right of the people and well regulated militia. In short, the evidence is therefore mutually reinforcing—and that’s without considering points addressed in my analysis that I don’t have space here to discuss.
the right (of the people) to bear arms
The corpus data provided no informative evidence of the use of the phrases right to bear arms or right to keep and bear arms, either with or without language identifying the right as belonging to “the people.” Putting aside a reference or two to heraldry and the right to bear a coat of arms, the only such uses in the data were from proposals for what became the Second Amendment and from provisions of state constitutions.
Relying on the text that was ultimately adopted as Second Amendment would obviously amount to begging the question. And the state provisions are inconclusive because in each such provision, bear arms was modified by a prepositional phrase that has no analogue in the Second Amendment:
bear arms for the defence of themselves and the state
bear arms, in defense of himself and the state
bear arms in defense of themselves and the State
It seems to me that it’s inappropriate to assume that the use of bear arms without any modification would have been understood in the same way as the use of the phrase as modified in the state provisions.
It appears from the corpus data, therefore, that right to bear arms was not in general use and therefore that the phrase had no preexisting conventionalized meaning. (That’s also true of right to keep and bear arms.)
Where does that leave us as to how the right to bear arms would have been understood when the Second Amendment was framed and ratified? I think we can confidently say that it would most likely have been understood as a right having to do with the militia, not as an individual right of self-defense unrelated to military service.
But that conclusion is rather vague about the way in which the right would have been seen as relating to the militia (and it doesn’t deal at all with the right to keep arms). In the analysis I completed in 2019, I said that the right to bear arms was most likely understood to protect a right to serve in the militia. But after further reading and reflection since then, I’m not fully satisfied with that conclusion. That’s not necessarily to say I think it’s wrong, but I suspect that at a minimum it’s an oversimplification.
The conclusion that there was a right to serve in the militia presupposes the existence of militias in which to serve. So perhaps the Second Amendment was understood at least in part as protecting the states’ ability to maintain militias. That suggestion will probably call to mind the “collective right” interpretation, which had largely gone out of favor by the time Heller was decided. While I’m not endorsing that interpretation, I tend to think that it deserves another look, especially considering the possible implications for the interpretation of keep arms.
In fact, much if not all existing Second Amendment scholarship is due for reexamination in light of the corpus evidence. To be more specific, what I think needs to be reexamined is any scholarship that interpreted bear arms as meaning ‘carry weapons’ (whether or not such carrying was thought to be associated with militia service). And that, in turn, probably encompasses a large percentage of Second Amendment scholarship—on both sides of the issue.
Of course, the necessary adjustments will pose a bigger problem for gun-rights advocates than for their opponents.
Source Note. For the text of the militia statutes I’ve referred to statutes, I relied on a 14-volume collection covering the period 1607-1789, which is entitled Military Obligation: The American Tradition and was published by the Selective Service System in 1947, [link].