That is an altogether presumptuous title, written with a smile. The case is Lockhart v. United States, 136 s. Ct. 958 (2016). It’s fascinating for the debate over conflicting canons of construction, the import of related statutes, and the value of legislative history. Think of it as a perfect vehicle for examining what seems to me the Court’s overreliance on textual methods of interpretation, and especially on certain canons.
In hundreds, if not thousands, of cases, courts have faced the kind of syntactic ambiguity that caused trouble in Lockhart. The solution does not typically lie in parsing and picking between textual canons. (The discourse on canons below is meant to make that point.) Courts must try to ground their decisions in something less mechanical when grappling with this recurring ambiguity. I offer my analysis in the form of an opinion by a self-appointed justice. Parts of the opinion will borrow from the two actual opinions — especially the dissent — without attribution. But the approach is radically different from either of them. Among other things, you’ll notice an uncommon candor and willingness to consider scholarly opinion (including — surprise — my own). Admittedly, the opinion would unsettle some interpretive pegs. If you wanted to read Lockhart at this point, that might help. The choice of examples on page 45 would then make more sense. But you don’t need to; the opinion should (naturally) explain itself.
So here goes a flight of fancy.
Justice Kimble delivered the Court’s opinion. Avondale Lockhart was first convicted under New York law for sexual abuse of his adult girlfriend. He was later indicted in the Eastern District of New York for child-pornography offenses under 18 U.S.C. § 2252(a) and pleaded guilty to one offense. His mandatory minimum sentence was increased under § 2252(b)(2) because of the earlier state conviction. The contested, confusing language from § 2252(b)(2) is this: “a prior conviction . . . under the laws of any State relating to aggravated sexual abuse, sexual abuse, or abusive sexual conduct involving a minor or ward.” We have here the all-too-common ambiguity caused by a trailing modifier: does involving a minor or ward modify all three items in the preceding series or just the last one? We agree with Lockhart that it could plausibly modify all three, and that because his earlier conviction involved an adult, his mandatory minimum sentence should not have been increased. We reverse the Second Circuit’s holding to the contrary.
To begin with, § 2252 is all about (as its codified title suggests) “the sexual exploitation of minors.” Decisions about its application should at least take into account that central concern. Section 2252(a) proscribes (very broadly) four activities involving child pornography:
Section 2252(b) then provides for sentencing, with enhancements (odd word that we have come to use) for a list of prior offenses. Under (b)(1), the enhancements are somewhat longer for violating (a)(1)–(3) than they are under (b)(2) for violating (a)(4).
At any rate, (b)(2) lists six categories of prior convictions for which enhancements are required. They are for convictions under:
This list is an assortment of sexually related crimes — pornography, obscenity, rape, sexual assault, sexual abuse, transporting for prostitution. The first one, the chapter we’re dealing with, involves crimes against children only. The next four, though, involve crimes against adults and children. That may point toward interpreting the last one, the state offenses, to include both — in other words, to read the trailing modifier involving a minor or ward as applying to the final item only. But this indicator is pretty weak: there’s no reason to think that Congress was seeking any kind of match between state and federal crimes involving prior sexual abuse. Indeed, the federal crimes include obscenity-related convictions under chapter 71 — mailing obscene matter, for instance — that no interpretation of our contested state-law provision would reach.
The government contends that the three items in our contested provision parallel three of the section titles in chapter 109A. Here they are side by side:
|“aggravated sexual abuse”||“Aggravated Sexual Abuse”(18 U.S.C. § 2241)|
|“sexual abuse”||“Sexual Abuse” (18 U.S.C. § 2242)|
|“abusive sexual conduct involving a minor or ward”||“Sexual Abuse of a Minor or Ward” (18 U.S.C. § 2243)|
|[none]||“Abusive Sexual Contact” (18 U.S.C. § 2244)|
Because of this correspondence, the government contends, the drafters must have been following 109A — meaning that they intended the first two items (on the left) to apply to crimes against any person, as the first two sections in chapter 109A (on the right) do.
But the correspondence is only partial. If the drafters were merely copying 109A, why do the third items differ, why does 109A include a fourth item, and what do we make of the difference between “abusive sexual conduct” on the left and “abusive sexual contact” in that fourth item from 109A? There’s no good explanation.
What’s more, if the drafters were duplicating 109A, then wouldn’t the substantive descriptions of “aggravated sexual abuse” and “sexual abuse” in 109A be somehow imported into whichever state-law crimes are covered? How would that work, exactly? Indeed, the government itself has rejected the idea that the state predicates mimic the crimes in 109A.1 But the same practical difficulty presents itself if — and this too can only be a guess — the drafters had in mind some “generic” sense of “aggravated sexual abuse” and “sexual abuse.” The federal courts would apparently have to decide in each instance whether the previous state crime fit that sense.
The government is essentially arguing that the state offenses “follow” 109A in a single respect, but not in any others — that is, in including sexual abuse of adults, but not in otherwise defining wrongful sexual conduct (whether concerning adults or children). It’s not a compelling argument.
The legislative history is almost as murky. The language at issue — aggravated sexual abuse, sexual abuse, or abusive sexual conduct involving a minor or ward — was first added to § 2252(b)(1) by the Child Pornography Prevention Act of 1996.2 It was added to § 2252(b)(2) two years later.3 The Department of Justice had recommended fixing the seeming discrepancy: after the 1996 change, (b)(1) provided enhanced penalties for those with “prior state convictions for child molestation,” but (b)(2) had no such provision for persons “who have prior convictions for child abuse.”4 The Department of Justice said that (b)(2) should also have an increased mandatory minimum for someone with a “prior conviction for sexual abuse of a minor.”5 And Congress delivered in the 1998 Act.
When Congress passed the 1996 Act, in which the disputed language first appeared, it was focused on child pornography. The 1996 Act was driven by technological advances in “the recording, creation . . . and transmission of visual images and depictions, particularly through the use of computers.”6 And it added, as 18 U.S.C. § 2256(8), a new definition of “child pornography” as involving a “visual depiction . . . produced by electronic, mechanical or other means” of various kinds of “sexually explicit conduct” involving minors. But the accompanying Senate Report barely discussed the enhancement language in question. It did refer in one place to “any State child abuse law.”7 Maybe this indicates that the state predicates must be limited to crimes against children. And maybe that indication is strengthened by the references to prior convictions for “child molestation” and “child abuse” and “sexual abuse of a minor” in the Department of Justice’s letter that prompted Congress to add the identical enhancement language to (b)(2).
Maybe. And yet Congress’s intense focus in those two Acts on sexual crimes against children makes it seem just as likely that Congress never thought about previous sexual crimes against adults (or adult wards). It would be perfectly natural in this context to say something like “And we’re enhancing the penalties for people with a previous conviction for child sexual abuse under state laws” — without reflecting one way or another on adult victims. So even if Congress did pluck the terms aggravated sexual abuse and sexual abuse from chapter 109A, we know only that previous sexual abuse of children has to be included. Congress’s concern was with protecting children. We don’t know whether that concern extended to enhancing penalties for someone who had sexually abused an adult.
One drafting point. Even without reconstructing (b)(2), Congress could have written: “any state law relating to aggravated sexual abuse or sexual abuse, to abusive sexual conduct involving a minor or ward, or to . . . .” Or (same meaning): “under any state law relating to abusive sexual conduct involving a minor or ward, to aggravated sexual abuse or sexual abuse, or to . . . .” Or (alternative meaning): “under any state law relating to aggravated sexual abuse or sexual abuse involving a minor or ward, to abusive sexual conduct involving a minor or ward, or to . . . .” The very clumsiness of the third fix, compared with the first two, may indicate, slightly, that the drafters expected the trailing modifier to apply across the board. If anything, the legislative history favors Lockhart. But it’s largely inconclusive.
We are faced with the inherent conflict between two canons (or rules, or doctrines, or principles, or maxims) of interpretation: the last antecedent canon as opposed to the series-qualifier canon. The first presumptively applies the trailing modifier to the last item in the series only; the second, to all the items. It’s time to reexamine them.
First, the last antecedent. The Court has applied that doctrine from our earliest decisions to our most recent.8 In Barnhart v. Thomas, we said that “a limiting clause or phrase . . . should ordinarily be read as modifying only the noun or phrase that it immediately follows . . . .”9 But we have never engaged the scholars who have heavily criticized the canon.10 Yes, we have acknowledged that it “can assuredly be overcome by other indicia of meaning.”11 The question is whether it deserves any interpretive weight at all. Because meaning in English often depends on placement, a reader’s first inclination is to link a modifier to the closest word or phrase. But first inclinations are no way to resolve ambiguity.
Consider the example we used in Barnhart to support the canon. Parents who are leaving for the weekend warn their teenage son: “You will be punished if you throw a party or engage in any other activity that damages the house.”12 We said: “If the son nevertheless throws a party and is caught, he should hardly be able to avoid punishment by arguing that the house was not damaged.”13 On reflection, the last antecedent may not have provided much of a footing for our interpretation. Most readers would be guided by the situ-ational (not verbal) context: no parents want their teenage son to throw a party while they’re gone. It’s intuition, or common sense, that provided the footing.
(Note, by the way, that the word other doesn’t seem to matter. Without it, the modifying phrase that damages the house surely wouldn’t apply to throw a party. If anything, other creates — rather than resolves — ambiguity.)
As for the conflicting canon, the series qualifier, its origins are less distinct in our opinions.14 We have twice described a “principle” that works the same way: “When several words are followed by a clause which is applicable as much to the first and other words as to the last, the natural construction of the language demands that the clause be read as applicable to all.”15 But this formulation is little more than a tautology.
It was Justice Antonin Scalia, our late colleague, and Bryan Garner who named and essentially formulated the seriesqualifier canon in their book Reading Law. 16 They made it hinge on the general nature of the series: “When there is a straightforward, parallel construction that involves all nouns or verbs in a series, a prepositive or postpositive modifier normally applies to the entire series.”17 It’s the “postpositive” part, of course, that conflicts with the last-antecedent canon.
So which canon trumps the other? Anybody could produce multiple variations on a single example to try to swing the debate. Suppose that you, the general manager of a majorleague baseball team, are describing your wish list for next year’s team. What does the trailing modifier apply to in the five bulleted sentences below (written in early 2017)?
These are difficult, debatable calls — and that’s the point. Other readers might disagree with some or all of them (except the first, no doubt). But every time you have modifiers in a series, either leading or trailing it, there’s a great risk of ambiguity — unless the drafters have been careful. And the resolution does not lie in merely picking between two canons but in considering the statute’s broad context, its purpose, the legislative history, sensible policy, and anything else that seems pertinent. Often, we’re thrown back on our own intuitive sense of the probable meaning. And often, the answer is just not clear, or even fairly clear.
Now, it does seem generally true that the simpler, shorter, and more parallel the series, the more likely it is that the modifier applies — or was intended to apply — across the board. That may be somewhat less true for trailing modifiers than for leading modifiers. Anyway, we should not preoccupy ourselves with trying to assess how straightforward and parallel the series in this case is: first, because we would probably disagree; and second, because even the simplest series can present uncertainty. It may well be that trailing modifiers should be treated as presumptively ambiguous.
Finally and more briefly, the surplusage canon — that we should “give effect, if possible, to every clause and word of a statute”21 — is also unavailing. If the modifier applies across the series, then “abusive sexual conduct involving a minor or ward” covers and makes redundant the first two offenses: “aggravated sexual abuse (involving a minor or ward)” and “sexual abuse (involving a minor or ward).” If the modifier applies only to the third offense, then the second one, “sexual abuse,” covers and makes redundant the first and third: “aggravated sexual abuse” and “abusive sexual conduct involving a minor or ward.” And this second reading would effectively scrub involving a minor or ward from the enhancement provision — contrary to the Act’s focus on the sexual exploitation of minors. In candor, we ought to acknowledge that the surplusage canon — like the last-antecedent canon — has been criticized as exceedingly weak.22 Neither one shines even a faint light here.
Only in the first bulleted example above — involving a veteran shortstop — were we fairly certain about the interpretation. In several others, we used the word probably. But how probable must a reading be for us to say that the language is unambiguous?
In 1904, we equated ambiguity with being “susceptible of two reasonable interpretations.”23 This is by far the prevailing definition in federal and state courts.24 A much stricter test — apparently used in just one or very few jurisdictions — is whether language is “equally susceptible” of more than one meaning.25 The Scalia–Garner book offers a third definition, one that’s in between but seemingly closer to the stricter test: “An uncertainty of meaning . . . that gives rise to any of two or more quite different but almost equally plausible interpretations.”26
Note the use of reasonable in the first definition and plausible in the third. Surely they mean the same thing in this context, or at least do not produce different results. At this point, one might be tempted to take a stab at quantifying. But even that would be fraught with disagreement. One federal judge reports that “if the interpretation is at least 65–35 clear, then I call it clear . . . .”27 Some of his colleagues, though, apply more of a 90–10 rule, while others apply a 55–45 rule.28 That’s quite a variation. Call it an uncertainty about an uncertainty.
So how would we put numbers to the three definitions?
While interesting as a diversion and perhaps marginally useful, this exercise does not get us very far. Legal interpretation, like the law itself, more or less depends on general standards — vague terms. (Indeed, when does a series qualify as “straightforward”? See above.) Our standard for ambiguity is “susceptible of two reasonable interpretations.” If ever a case met that definition, this one does.
When it comes to applying the rule of lenity in criminal cases, some of our pronouncements seem to have further complicated the meaning of ambiguous.
On the one hand, we have used an intensifier that might suggest some kind of heightened standard: “To invoke the rule, we must conclude that there is “a ‘grievous ambiguity or uncertainty’ in the statute.”29 We make no distinction between ambiguity and grievous ambiguity.
On the other hand, we might on occasion have suggested a diminished standard: “[I]f our recourse to traditional tools of statutory construction leaves any doubt about the meaning . . . , we . . . invoke the rule that ‘ambiguity concerning the ambit of criminal statutes should be resolved in favor of lenity.’”30 The word any in any doubt was too strong, too broad. We have since more accurately stated that the rule kicks in when, “after all legitimate tools of interpretation have been exhausted, ‘a reasonable doubt persists’ regarding whether Congress has made the defendant’s conduct a federal crime.”31 A reasonable doubt, not any doubt or a possible doubt.
We also seem to have developed two equivalent approaches. (1) A criminal statute is ambiguous if it is susceptible of two reasonable interpretations. If it’s ambiguous, the rule of lenity applies. (2) The rule of lenity applies if there’s a reasonable doubt about whether a statute makes the defendant’s conduct a crime. Approach (2) prompts a fair question: do we need to interpose the concept of ambiguity, as (1) does? Perhaps scholars will shed some light on that in days and articles to come.
In any event, all the tools of interpretation have been exhausted in this case, and we are still left with two reasonable, or plausible, interpretations — and thus a reasonable doubt. The government’s argument from the structure of the statute is weak. Lockhart’s argument from legislative history is not much stronger. The textual canons of construction are no help. We therefore invoke the rule of lenity. The judgment of the U.S. Court of Appeals is reversed, and the case is remanded for further proceedings.