[An introduction and guide to my series of posts “Corpora and the Second Amendment” is available here.]
Before I get down to the business of discussing the corpus data and its implications for the Supreme Court’s analysis in Heller, I want to say a few things about what this series of posts will and won’t be about, I want to offer some caveats, and I want to outline the sequence that the posts will follow.
What the posts will and won’t be about
These posts are going to focus on the meaning of the phrase keep and bear arms and on the Court’s analysis of that phrase. I won’t be talking about the other parts of the Second Amendment (a well-regulated militia, the security of a free state, the right of the people, and infringed).
The discussion will concentrate on linguistic issues rather legal issues. I won’t be talking about whether the Court’s holding in Heller is correct. I will, however, talk about what my linguistic analysis means for Heller‘s conclusion that the Second Amendment’s text is unambiguous and therefore that the prefatory clause plays no role in the amendment’s interpretation.
[Update (Aug. 5, 2019)]: As I reread the two paragraphs above, now that this series is on the verge of being completed, I’m not sure that I’ve strictly adhered to the limits that I described when I wrote this post, in June 2018. In fact, I know that I went beyond one of them: despite what I said here, I did deal to some extent with the right of the people. But rather than try to revise this introductory post so that it conforms to how the series ultimately turned out, I’ll leave it as is, with only the addition of this disclaimer.]
I will be working within the interpretive framework on which Heller‘s analysis of keep and bear arms is based. Under that approach, the goal is to determine how one would expect the Second Amendment’s text to have been understood by members of the public at the time it was framed and ratified. This is known variously as “New Originalism” and “original-public-meaning originalism,” and it differs from the variety of originalism that it has largely replaced—“original-intent originalism”—in that the former focuses on how the text is likely to have been understood, while the latter focuses on what the framers intended. In this respect, original-public-meaning originalism is similar to textualism. However, I don’t think that that prevents interpreters from taking account of whatever inferences late-18th century readers could reasonably have drawn from the text as part of the ordinary process of reading and understanding it. (For further discussion, you can read this post.)
The fact that I’ll be following originalist methodology shouldn’t be taken as indicating that I’m an originalist—or any kind of an -ist. As I’ve previously said, I’m ecumenically agnostic about the various theories of interpretation, and in any case corpus linguistics is compatible with any of the recognized theories. In applying original-public-meaning originalism, I am merely trying to examine evaluate Heller through the lens of the interpretive approach that was followed by Heller itself.
The caveats I want to raise all involve issues relating in one way or another to the corpus data that I’ll be discussing.
The first caveat concerns the selection of the texts that comprise the corpora: to what extent is the data representative of American English of the late 18th century? The question of representativeness isn’t unique to the BYU Law corpora; it is relevant to all corpora. For that matter, it is also relevant to the data that the Supreme Court relied on in Heller, the scope of which pales in comparison to the contents of the corpora here. That suggests to me that at a minimum, the corpus data will be more reliable than what was cited in Heller.
But evaluating corpus design is not something about which I have any particular knowledge expertise. Therefore, I can’t do much more than flag the issue, and leave it to be addressed by someone who is qualified to do so. And that person, whoever they are will probably need to have more information about the makeup of the corpora than has yet been made public. Hopefully the release of such information is on the BYU Law School’s to-do list.
Another caveat relates to the corpus data that I’ve made available for downloading (and the additional data that I will make available in the future). And as caveats go, it’s a pretty obvious one: the data that results from the searches that I do is by definition a function of the way in which the searches are framed. Framing a search more broadly than I did might very well turn up additional relevant data. But it would do so at the expense of also turning up additional irrelevant data that would have to be waded through. So as with many kinds of research, the search strategy that is chosen will often require tradeoffs between competing considerations.
While I think my choices have struck a sensible balance, some people might disagree. If you’re one of those people, I’d encourage you to discuss your disagreement in the comments. If I think it’s appropriate, I might supplement the data or explain why I disagree. And of course, you’re free to run your own search.
The plan for the upcoming posts
In my next post, I will summarize and comment on the relevant parts of Justice Scalia’s majority opinion in Heller. After that, I’ll turn to the specifics of keep and bear arms, discussing both the corpus data and various other issues (including whether late-18th century Quakers were vegetarians) as we go along. I’ll have at least one post on each of the following topics, in the following unsurprising order: keep, bear, arms, and keep and bear arms. At appropriate points, I will discuss relevant aspects of the approach to word meaning that I am following, which will be different from the dictionary-based approach that most people are familiar with. I’ll try to integrate that with the discussion of the text and the corpus data, but I might end up breaking some of it out into a separate post.
[Cross-posted on Language Log.]
Why? The phrase “keep and bear arms” is least likely to get good independent coverage in a corpus. So why not look at “well-regulated militia”? It is much shorter.