Textualists believe in using legislative history to interpret statutes—even when, if not for a statute's legislative history, they would otherwise find the statute's text clear. What grounds could I have for making such a provocative remark? It's famously true, of course, that textualists reject the notion of legislative history as authoritative gloss. That's because textualists believe a statute's legal meaning is what its language means in some objective sense, not what Congress intended to convey by it. But textualists have always granted that legislative history can help us work out what statutory language means, if only because legislative history, like any other writing or speech, can serve as evidence of how particular words or phrases are used, and thus what those words or phrases mean. Here, for example, is Justice Scalia discussing the point in a published dialogue with John Manning five years ago:
And by the way, I don't object to all uses of legislative history. If you want to use it just to show that a word could bear a particular meaning--if you want to bring forward floor debate to show that a word is sometimes used in a certain sense--that's okay. I don't mind using legislative history just to show that a word could mean a certain thing. We are trying to ascertain how a reasonable person uses language, and the way legislators use language is some evidence of that, though perhaps not as persuasive evidence as a dictionary. That is using legislative history as (mildly) informative rather than authoritative: “the word can mean this because people sometimes use it that way, as the legislative debate shows,” rather than “the word must mean this because that is what the drafters said it meant.”
There's an interesting ambiguity here. Scalia's clear that he's not persuaded by arguments that statutory language must mean this because that's what the drafters said it meant. It's unclear, though, whether he's open to arguments that statutory language can mean this because that's what the drafters said it meant to them, or whether the only sorts of legislative history he'll consider for this purpose are statements that incidentally use a word or phrase in a certain way, rather than defining it or glossing it.
I think it's likely that Scalia would at least be reluctant to use gloss as evidence of linguistic meaning; Scalia seemed to distrust legislative gloss apart from his concern that it not be deemed controlling. Earlier in his response to Manning's question on legislative history, he complains about legislative history that "simply declares what the committee or sponsor intends a word or phrase to mean: 'Subsection B means this or that.' That statement is meant to be authoritative; its one and only function is to tell us how that committee or sponsor wants the bill to be interpreted. When judges attribute that intention to Congress as a whole, they are not ascertaining meaning . . . ." However, if legislative history can be evidence of what a word in a statute can mean, what's better evidence, as far as legislative history goes, than a definitive statement by that statute's drafters on what they think that word means in that statute? A floor speech in which a Senator uses a statutory term to mean something or another only shows what the term can mean in a context that may bear little resemblance to the context in which it's used in the statute; we may also be generally suspect of informal usage as a proxy for meaning in the statute's relatively formal and technical context. On the other hand, a statement by the drafters that some snippet of statutory language means something to them would seem much stronger evidence of what that language at least can mean to reasonable people in the context of the statute itself.
Bracketing for the moment Scalia's possible hostility to using legislative gloss as evidence of linguistic meaning, it's at least clear that Scalia was open to using some legislative history as evidence of what statutory language can mean, but not open to using legislative history as authoritative evidence of what statutory language does mean. This use of legislative history to suggest linguistic possibilities, but not to resolve linguistic debates, has an intriguing corollary; for Scalia at least, legislative history is much more probative when a statute seems clear than when it's ambiguous. When Scalia already knew from other sources that a statute was ambiguous, legislative history had no work to do for him; it could at most redundantly confirm that a statute could mean what a dictionary or his own understanding of the language already told him it could mean. But when a statute seemed clear, legislative history could productively dispel that appearance of clarity and show him that a "word can mean this because people sometimes use it that way."
This seems counterintuitive given textualists' devotion to the plain meaning rule, but the internal textualist logic is perfectly sound. As textualists don't think a statute means what Congress intended, the usual way people use legislative history—as a solvent of ambiguity if the statute's text isn't clear—makes no sense for textualists. There's just no reason for textualists to say that if a statute's ambiguous, it means what "Congress" (really, certain involved committees or members) said it meant. But textualists are interested in what statutory language means, and absent argument that Congress is so linguistically incompetent that its usage is worthless evidence of linguistic meaning, it makes no sense to have a rule by which legislative history is disregarded so long as a statute seems clear prior to considering evidence of alternative usage in legislative history. That would be tantamount to saying that courts should decide whether a statute's clear on less than all the relevant evidence. Indeed, on Scalia's understanding of legislative history's uses, it would seem that the only time when legislative history matters, at least in any dispositive way, is when a statute would seem clear if not for legislative history.
A famous opinion by Judge Easterbrook confirms that textualists, or at least the judiciary's two leading textualists, see legislative history's use as a dissolvent of seeming clarity rather than a solvent of ambiguity. Matter of Sinclair, perhaps the finest statement of textualist theory a court has ever produced (though it's hardly immune from criticism) was written at a time,1989, when textualism was far from the dominant modality of statutory interpretation. In explaining why he was taking a textualist approach to legislative history—specifically, legislative history that flatly stated a bill had limited retroactive application when the statute in question just as flatly stated its "amendments . . . shall not apply with respect to cases commenced . . . before the effective date of this Act"—Judge Easterbrook felt he had to reconcile that approach with then-still-vital cases that said legislative history bore on statutory interpretation even when a statute was plain. His solution was to say that legislative history could bear on whether a statute was plain—that evidence of alternative usage in legislative history can dissolve a seeming clarity:
What's a court to do? The answer lies in distinguishing among uses of legislative history. An unadorned "plain meaning" approach to interpretation supposes that words have meanings divorced from their contexts — linguistic, structural, functional, social, historical. Language is a process of communication that works only when authors and readers share a set of rules and meanings. What "clearly" means one thing to a reader unacquainted with the circumstances of the utterance — including social conventions prevailing at the time of drafting — may mean something else to a reader with a different background. Legislation speaks across the decades, during which legal institutions and linguistic conventions change. To decode words one must frequently reconstruct the legal and political culture of the drafters. Legislative history may be invaluable in revealing the setting of the enactment and the assumptions its authors entertained about how their words would be understood. It may show, too, that words with a denotation "clear" to an outsider are terms of art, with an equally "clear" but different meaning to an insider [emphasis added]. . . These we take to be the points of cases such as American Trucking holding that judges may learn from the legislative history even when the text is "clear". Clarity depends on context, which legislative history may illuminate. The process is objective; the search is not for the contents of the authors' heads but for the rules of language they used.
As with Scalia's remarks to Manning, one may wonder exactly how the two halves of this dichotomy are to work. Once one grants that legislative history may show that language that clearly means one thing to a court can or even does mean something else (unlike Scalia, Easterbrook seems open in this passage to arguments that legislative history may show what a statute does mean, not just what it may possibly mean), how can legislative history ever "show an 'intent' at variance with the meaning of the text"? If a committee report may show that language that appears to plainly mean x has "a strikingly different meaning" of y, can't any legislative history that seems "to show an 'intent' at variance with the meaning of the text" be recharacterized as evidence of the "strikingly different meaning" that text had to the "insider" writing it?Quite different is the claim that legislative intent is the basis of interpretation, that the text of the law is simply evidence of the real rule. . . . The "plain meaning" rule . . . rests not on a silly belief that texts have timeless meanings divorced from their many contexts, not on the assumption that what is plain to one reader must be clear to any other (and identical to the plan of the writer), but on the constitutional allocation of powers. The political branches adopt texts through prescribed procedures; what ensues is the law. Legislative history may show the meaning of the texts — may show, indeed, that a text "plain" at first reading has a strikingly different meaning — but may not be used to show an "intent" at variance with the meaning of the text. [Emphasis added.]
I think Easterbrook would say (and in Sinclair, at least begins to suggest) two things in response. The first and perhaps less persuasive is that some language's meaning is just too plain for legislative history to show it can mean something else. If Congress writes that a law "shall not apply to cases commenced before the effective date of this Act," even legislative history that claimed that that precise language meant that courts had discretion to apply the law retroactively would be hard to take as a gloss of what that language truly meant to the legislative history's authors. Rather, we might suspect that they were attempting to manufacture ambiguity on an issue they lost. Second, not all or even most legislative history will speak directly to the question of what statutory language means. The legislative history in Sinclair didn't actually regard the statute's effective-date clause; it glossed another section of the statute, and it seems unlikely that the drafters of that history were even aware of the effective-date clause's existence. In other cases, legislative history may generally describe objectives that members of Congress intended but didn't achieve in every (or any) detail. It's in this kind of sense that we should take Easterbrook's somewhat cryptic concluding remark in Sinclair that "[n]o legislative history suggests any other meaning [than the apparently plain one]. The committee report suggests, at best, a different intent."
II. Legislative History in SW General
a. The statute
We can now turn to the Court's opinion in NLRB v. SW General. SW General was a case about whether a provision in the Federal Vacancies Reform Act (FVRA) that prohibited nominees to a position from serving in that position in an acting capacity while they wait to be confirmed—unless they had previously served as the first assistant to that position for at least ninety days—applied to (a) all types of acting officers, or (b) only those who became acting officer by automatic succession by virtue of being the vacant office's first assistant.
The FVRA provides for three types of acting officers in subsections (a)(1), (a)(2), and (a)(3) of FVRA's Section 3345. The first, under (a)(1), are first assistants to an office, who succeed automatically to an acting role in the case of a vacancy in that office unless the President acts under subsections (a)(2) or (a)(3). Under (a)(2), the President can direct any Senate-confirmed executive-branch official to serve in an acting capacity in any vacant office. And under (a)(3), the President can designate any officer of an agency who's been in the agency for ninety days, and is paid at a GS-15 rate or greater, to serve in an acting capacity in any vacancy in his agency.
In SW General, the President used his (a)(3) powers to elevate Lafe Solomon, a senior official at the NLRB of ten years' standing, to serve as the NLRB's acting General Counsel. He then nominated Solomon to serve as the NLRB's GC permanently—triggering, potentially, the prohibition in 3345(b)(1) that bars nominees to an office from acting in that office unless they served as first assistant to the office for ninety days. Solomon had never been the NLRB GC's first assistant, so it followed that he couldn't serve as the acting GC during his confirmation process—unless 3345(b)(1)'s prohibition didn't apply to (a)(3) acting officers, as the NLRB argued.
The NLRB's textual basis for arguing that 3345(b)(1) didn't apply to (a)(3) officers was rather thin. 3345(b)(1) begins: "Notwithstanding subsection (a)(1), a person may not serve as an acting officer for an office under this section. . . ." The NLRB argued that "[n]otwithstanding subsection (a)(1)" meant that the prohibition only applied to (a)(1) acting officers; the private parties challenging Solomon's official acts as void argued that the reference to (a)(1) merely alleviated a uniquely strong tension between (a)(1) and (b)(1), since (a)(1) mandated that first assistants shall serve as acting officers and (b)(1) mandates that they shall not (absent sufficient first-assistant service) once nominated to their temporary office. The private parties also argued that (i) notwithstanding simply doesn't mean "notwithstanding and as to," as the NLRB would have it mean, (ii) that (b)(1)'s bar on any "person" serving "under this section" made it plain that (b)(1) applied to all acting officers appointed under any part of 3345, not just (a)(1) acting officers who succeeded to their posts under (a)(1), and (iii) that the NLRB's interpretation created surplusage in multiple parts of the statute.
All that is quite compelling; perhaps most compellingly, "notwithstanding subsection (a)(1)" just doesn't seem to create the textual daylight the NLRB argued it did. As Justice Kagan reasoned at oral argument, suppose "I'm talking to my waiter, and I place three orders. I say, number 1, I'll have the house salad. Number two, I'll have the steak. Number three, I'll have the fruit cup. And then I tell the waiter, notwithstanding order number three, I can't eat anything with strawberries. So on [the NLRB's] theory, the waiter could bring me a house salad with strawberries in it, and that seems to me a quite odd interpretation of what's a pretty clear instruction: No strawberries." Quite so. Notwithstanding order number three just means, well, notwithstanding order number three, or despite order number three (which might seem to call for strawberries), not "but as to order number three, and only order number three." And the same seems true of "notwithstanding subsection (a)(1)"; it just means, in any ordinary sense of the phrase, "despite subsection (a)(1) [which the following might seem to present a difficult conflict with if not for this indication that this subsection shall trump (a)(1)], the following is the rule"—not "but as to, and just as to, (a)(1), the following is the rule."
And yet I hesitate. For as clear as it is to me, virtually any neutral lawyer who's thought about this case, and all the lexicographers that "despite," not "but as to," is the prescriptively correct usage of "notwithstanding," it wouldn't at all surprise me if a substantial fraction of high-school-educated or even college-educated Americans would read "notwithstanding subsection (a)(1)" to signal an (a)(1)-specific exception, or if the same fraction uses notwithstanding to mean "but as to." Perhaps widespread misusage (but who decides what's misusage?) doesn't ambiguity make; perhaps the only understandings that matter, for purposes of a statute addressed to acting officers in Senate-confirmed positions, are educated legal understandings, though here it's interesting to note that the Office of Legal Counsel and the executive branch had always read (b)(1) the NLRB's way. These, however, are rather arguable points; what I should really like before pronouncing the statute clear is some confirmation that my and others' relatively sophisticated understandings of "notwithstanding" fully capture the range of meanings that reasonable readers of English could suppose the reasonable writers of English in Congress had in mind when they used the word.
b. The legislative history
It is here where the legislative history has some value. It seems that Senator Fred Thompson—successful Washington attorney, player of attorneys on Law and Order—drafted much of the FVRA, and in particular the reference to (a)(1) in (b)(1). Explaining to his colleagues, in great and highly accurate detail, the many changes to the bill that had been negotiated since it last saw the Senate floor, Thompson said this of (b)(1) as it reads today:
The 180 day period in 3345(b) governing the length of service prior to the onset of the vacancy that the first assistant must satisfy to be eligible to serve as the acting officer is reduced to 90 days. Under 3345(b)(1), the revised reference to 3345(a)(1) means that this subsection applies only when the acting officer is the first assistant, and not when the acting officer is designated by the President pursuant to 3345(a)(2) or 3345(a)(3). The 90 day service requirement is inapplicable to a first assistant who has already received Senate confirmation to serve in that position [this is an accurate synopsis of 3345(b)(2)].What "revised reference to 3345(a)(1)" was Thompson talking about? The phrase "notwithstanding subsection (a)(1)." It's clear that Thompson, the "author" or proponent of that language, thought it meant that what followed applied only to (a)(1), not (a)(2) or (a)(3). To put the point even more strongly, it's clear that Thompson, who evidently only wanted (b)(1) to be an exception as to (a)(1), believed he could make that point by saying that (b)(1) applied "notwithstanding" (a)(1). This might not earn Thompson or his staffers a good grade for draftsmanship, but it's what an educated attorney of the time the FVRA was drafted thought notwithstanding meant in the context in which the Court was called upon to interpret it.
c. The Court's opinion
What does Chief Justice Roberts' scrupulously textualist opinion make of this legislative history? Nothing. In the first place, he tells us that the "ordinary meaning" of notwithstanding is "in spite of," citing a pair of dictionaries and Scalia and Garner's authoritative guidance. It's used to tell us "which provision prevails in the event of a clash," not to cabin the overriding effect of a provision to those it specifically negates. What of the fact that Congress strangely only mentioned (a)(1) in the notwithstanding clause, rather than (a)(2) and (a)(3), with which (b)(1) also clashes by cutting short acting officers' service? No matter; after all, if a hypothetical radio station were to awkwardly and rather unidiomatically say that "[w]e play your favorite hits from the '60s, '70s and '80s[; n]otwithstanding the fact that we play hits from the '60s, we do not play music by British bands," you wouldn't think they do play music by British '70s bands. Rather, you'd think they were particularly addressing the assumption that they would play music by The Beatles and Rolling Stones given that they play '60s hits, but not actually limiting their no-British-band rule to '60s hits. So too it follows that "notwithstanding (a)(1)" doesn't mean "but as to (a)(1)," but only "in spite of (a)(1) (implied: with which the following especially conflicts)." After some further remarks on various superfluities posed by the NLRB's reading, the statute is pronounced clear.
Then, and only then, Roberts comes to the statute's legislative history, including Thompson's understanding of his own use of "notwithstanding" to mean that (b)(1) only restricted (a)(1). As an initial matter, he dismisses all of it, writing that "[t]he text is clear, so we need not consider this extra-textual evidence." Eventually turning to Thompson, he points out that Senator Byrd—who offered a much less detailed exegesis of the revised FVRA that didn't specifically engage with (b)(1)'s cross-reference to (a)(1), and apparently played a lesser role in drafting the amendment in question—said after Thompson spoke that under (b)(1) any acting officer couldn't serve once nominated to his position absent ninety days of first-assistant service. (It's not quite clear that Byrd was saying this.) This disagreement between Thompson and Byrd, Roberts scoffs, "is a good example of why floor statements by individual legislators rank among the least illuminating forms of legislative history," which is sort of like saying that dictionaries aren't illuminating forms of interpretive guidance because they contain multiple definitions of the same words. And that's all SW General has to say about the FVRA's drafter's understanding of "notwithstanding."
d. My critique (finally)
Whatever the weight of Thompson's statement, a point on which I'm agnostic, it should be clear that Roberts' dismissal of it rests on a confusion about the role legislative history has to play in textualist interpretation. (This could be because Roberts, however many textualist opinions he produces, is the farthest thing from a textualist that the judiciary has to offer these days.) For textualists, legislative history isn't extra-textual evidence that can decide or even help decide the intended meaning of unclear texts; Thompson's statement wouldn't be any more relevant if dictionaries showed that notwithstanding could mean what he thought it meant. A textualist will be skeptical that Thompson's intent was necessarily Congress's intent, and thoroughly disinterested, in any event, in what Congress's actual intent was. So it doesn't matter if Thompson and Byrd disagreed on what they meant the statute to mean, because a textualist doesn't care what they actually meant.
Rather, to a rigorous textualist, legislative history is one of many sorts of secondary evidence, and not a particularly privileged one, of what language in the text can mean. And because legislative history will only be probative when other, more conventional and authoritative sorts of evidence of meaning and usage (like dictionaries) don't already show an ambiguity, legislative history only especially matters to textualists when the text otherwise seems clear. So we cannot pronounce the FVRA clear before considering that its drafter read the key word in the case differently than the dictionaries do; nor, for that matter, can we pronounce the FVRA clear before considering that the Office of Legal Counsel issued contemporaneous post-enactment guidance reading the reference in (b)(1) to (a)(1) to limit (b)(1)'s prohibition to (a)(1). We must at least consider the possibility that the text isn't clear because some people in the community of federal officials and lawyers to whom the statute was addressed (though not Senator Byrd) understood it, at the time it was enacted, to mean something other than what it seems to clearly mean.
Now, there may be, of course, a variety of textualist arguments that Thompson's particular statement doesn't matter even if legislative history can shed light on usage in theory. Let's consider a few of these. They'll all, as we'll see, track Scalia and Easterbrook's caveats on textualist uses of legislative history.
First, a textualist professor and friend has suggested to me, along the lines of Easterbrook's distinction in Sinclair between legislative history that reveals "a different intent" and legislative history that truly goes to textual meaning, that Thompson wasn't really talking about what the text meant, but just what he hoped the bill to achieve. I deem this a thoroughly implausible reading of Thompson's remarks that's nothing more than a textualist's characteristically motivated effort to avoid engaging at all with legislative history, no matter how relevant to textual meaning it may be. Thompson says that "[u]nder 3345(b)(1), the revised reference to 3345(a)(1)," that is, the phrase "notwithstanding subsection (a)(1)," "means that this subsection [3345(b)(1)] applies only when the acting officer is the first assistant [under (a)(1)], and not when the acting officer is designated by the President pursuant to 3345(a)(2) or 3345(a)(3)." Legislative history doesn't get more anchored to the text than this remarkably granular statement about what a revised cross-reference in one subsection to another "means." One might as well say that the Court's opinion in SW General itself isn't really about what the language in 3345(b)(1) means.
Second, the same professor somewhat more reasonably suggests, along the lines of Scalia's distinction between legislative history that says what a statute means and legislative history that shows how a word is used, that perhaps Senator Thompson was just prevaricating about what he thought the language meant in hopes that courts would read an idea he couldn't get into the bill's text into the bill. Here too, we seem to be fishing for reasons to avoid confronting legislative history. My friend's supposition isn't consistent with the FVRA's Clinton-era history, in which Senator Thompson and other Republican sponsors who wanted to limit President Clinton's acting-officer manipulations had to make deals with Democratic Senators to give the President more flexibility in order to get the bill past a filibuster. On my friend's supposition, Thompson would have been pretending that the FVRA gave Clinton more leeway to put his nominees into acting positions than his negotiating partners would allow. It also isn't consistent with how careful and accurate the rest of his remarks are, and one also has to wonder why no one would correct Thompson if he were really lying about getting an idea into the bill that his colleagues fought to keep out.
That said, there's no doubt that remarks of this kind can be self-serving, while a Senator's casual misuse of "notwithstanding" in a floor speech to mean "but as to" is much less likely to be a connivance to convince courts that "notwithstanding" had that alternative meaning in 1995. Of this concern I would say that absent evidence that when a member of Congress says to his colleagues that some statutory language has a meaning a dictionary doesn't support, he's usually trying to fool them or a court—evidence which I really don't see—we shouldn't exclude this sort of evidence of usage. We can discount for the risk that interpretive gloss in legislative history is disingenuous without discarding it altogether. Besides, while the risk of disingenuousness is greater with interpretive gloss of statutory language than casual use of some words that appear in a statute, the value of gloss, if sincere, seems much greater. "Notwithstanding" might mean something quite different in the phrase "notwithstanding subsection (a)(1)," and indeed in the context of the FVRA more broadly, than it means in hypotheticals like "notwithstanding order number three" or "notwithstanding the fact that we play hits from the 60s." We should be, I think, quite reluctant to discard evidence of how drafters of a phrase in a statute contemporaneously understood it. Conversely, we should be more humble than we often are about the epistemic value of homespun hypothetical usages that differ from the statutes we're interpreting in every possible way other than sharing a couple words and maybe some similar syntax.
Third, another textualist professor has suggested to me that Thompson's apparent understanding of notwithstanding might not be a standard use of the word, and that we can only give statutory language standard meanings. This is perhaps anticipated by Easterbrook's claim in Sinclair that there was simply only one way to read "shall not apply to cases commenced before the effective date of this Act," no matter what the legislative history might say. I'm more or less receptive to this proposal depending on what's meant by standard. If by standard we mean prescriptively acceptable, as some people will, I don't see how we can rule out usages that lots of people use as interpretive possibilities just because we're taught not to use them (which is usually because they were much less common at some time in the past). If Congress wants to use "irregardless," I can't imagine a justification for interpreting the word as gibberish, rather than what anyone would understand it was trying to say, because irregardless is bad English.
If by standard, however, we merely mean non-idiosyncratic, I do think a very strong case can be made that a statute isn't ambiguous just because one person—even if he were the statute's drafter—thinks it means something that no one else does. The more, however, evidence of use or understanding shows that Thompson wasn't the only one who read "notwithstanding" the way he did at the time the FVRA was enacted, the more we must deem his understanding "standard," even if irregular or relatively uncommon.
Finally, because people accustomed to non-textualist uses of legislative history will reflexively assume I'm suggesting one, nothing I've said is meant to suggest that we should use Thompson's statement to close debate on what "notwithstanding" meant in the FVRA, or even use it as evidence of any kind or weight of what Congress intended the FVRA to mean and thus what it does mean. I only suggest that Thompson's statement, along with the OLC's contemporaneous understanding, might show that the FVRA isn't as clear as it initially seems, and that we must consult these extra-textual sources of evidence of usage before we deem the statute clear, just as the Court consulted the extra-textual evidence of the dictionary, Scalia and Garner, and its intuitions about the meaning of hypothetical uses of "notwithstanding" in connection with '60s hits and fruit cups before deeming the FVRA clear. I'll even grant that the dictionary has more weight than the legislative history, though I tend to give the legislative history more weight than the fruit cups.