Introduction
In Bostock v. Clayton County, the Supreme Court held that Title VII of the Civil Rights Act of 1964 prohibits discrimination based on sexual orientation and gender identity, thereby delivering an important victory for LGBTQ+ persons in their continuing struggle to be treated with equal regard in all areas of life.1 A striking feature of the case, and one reason why it has been so widely discussed, is that all three opinions—the majority and two dissents—professed to apply a textualist theory of statutory interpretation.2 In particular, all three opinions took for granted that courts should enforce a statute’s ordinary meaning at the time of enactment.3 No competing theory of statutory interpretation was even on the table.
Going forward, we can expect textualism to play an increasingly prominent role in how courts resolve questions of statutory interpretation. So, it is worth asking what textualism instructs courts to do and whether courts should do as textualism instructs. A recent Article by Professors William N. Eskridge, Brian G. Slocum, and Stefan Th. Gries attempts to answer both of those questions.4 It contends that there were multiple versions of textualism on display in the Justices’ opinions in Bostock and that none of those versions is ultimately defensible.5 This is a long and rich Article by distinguished scholars, and I agree with much of what they say. Yet I also think that their characterization of and objections to textualism miss the mark.
In this Essay, I argue (i) that the versions of textualism that Eskridge, Slocum, and Gries criticize are not really textualism; (ii) that their examples of “societal dynamism” do not put any pressure on textualism properly understood; and (iii) that their corpus-linguistics analysis of the word “sex” would not persuade any textualist to adopt their preferred interpretation of Title VII. I am not a dyed-in-the-wool textualist myself: while frequently sympathetic to textualism, I doubt that judges ought to employ it in every case. Still, my sense is that many commentators are unduly dismissive of textualism—tending to criticize strawman versions of it rather than the genuine article—and my goal is to push back against that tendency here.6
Below, Part I attempts to lay out more clearly what textualism claims. Parts II and III argue that the “compositional” and “extensional” versions of textualism that Eskridge, Slocum, and Gries criticize are not textualism and, indeed, not positions that any mainstream legal interpreter today claims to hold. Part IV answers the authors’ objections to textualism based on so-called societal dynamism. Finally, Part V turns to the authors’ corpus-linguistics analysis of “sex.”
I. What Does Textualism Claim?
Textualism asks courts to enforce the most plausible interpretation of a statute’s objective asserted content, i.e., what a reasonable reader would have been most likely to infer that a legislature intended to assert by a statute in the context in which it was enacted.7 Let me explain what I mean. To start, we can distinguish semantic from asserted contents. Semantic content is a property of sentences in the abstract. The semantic content of a sentence is entirely a function of the conventional meaning of its words and the rules of syntax.8 Asserted content, by contrast, is a property of uses, or utterances, of sentences.9 The asserted content of an utterance is the complete proposition that a speaker uses that utterance to assert on a particular occasion.
We can conceive of asserted content in two ways, either subjectively or objectively.10 An utterance’s subjective asserted content (or “speaker meaning”) is the complete proposition that the speaker intends to assert by that utterance. By contrast, an utterance’s objective asserted content is the complete proposition that a reasonable hearer (or reader) would infer that the speaker intends to assert by that utterance. Objective asserted content is a function of not only (i) the semantic content of the sentence uttered but also (ii) beliefs a reasonable hearer would have about the context in which the speaker is speaking. When textualists speak of “ordinary meaning,” they might be referring to objective asserted content.11
Notably, the semantic content of a sentence often underdetermines the objective asserted content of any utterance of that sentence.12 While semantic content always constrains objective asserted content (e.g., there is no context in which uttering “I live in Ithaca” would objectively assert in standard English that it is raining in Chicago), semantic content often fails to uniquely determine which proposition an utterance objectively asserts. In that event, it falls to a reasonable hearer’s contextual beliefs to determine which of multiple semantically permissible propositions the utterance does objectively assert. Those contextual beliefs may include, for instance, beliefs about the speaker’s purpose in speaking, common usage of words or phrases, or commonly known features of the world.
When a reasonable hearer’s contextual beliefs would lead him or her to infer that a speaker intends to assert something above and beyond the bare semantic content of the sentence uttered, philosophers of language say that the utterance’s objective asserted content is “pragmatically enriched.”13 Much of what we assert in everyday conversation is pragmatically enriched to some extent. Take the following sentences:
(1) I haven’t had any breakfast.
(2) It’s sunny.
(3) Everyone should arrive by six.
In most contexts, a reasonable hearer would assume that a speaker who utters (1) has had breakfast at some point and so would infer that the speaker must intend to assert that he or she has not had any breakfast yet today. A reasonable hearer would also assume that it is not sunny always and everywhere in the world and so would infer that a speaker who utters (2) must intend to assert that it is sunny at a specific time and place (e.g., right now in Ithaca). And a reasonable hearer would believe that a speaker who utters (3) must intend to assert something about a certain domain of people (e.g., those invited to a certain party) and, moreover, must intend to assert that those people should arrive by six o’clock in the evening (when most parties occur).
Since legislatures tend to speak more explicitly than we do in everyday speech, pragmatic enrichment of legislative speech is usually subtler than in the foregoing examples. Still, one way in which context sometimes pragmatically enriches what a statute objectively asserts is by fixing the operative sense of a polysemous word. Whereas a word is lexically ambiguous if it has two conventional meanings, a word is polysemous if it has multiple related senses of the same conventional meaning.14 To use the classic example, the word “bank” is lexically ambiguous because it can refer to a financial institution or to the border of a river (two conventional meanings). Yet “bank” is also polysemous because it can refer to a banking company or to the building in which such a company is housed (two related senses of the same financial institution meaning).
Consider Andrus v. Charlestone Stone Products Company.15 The question there was whether a federal mining statute providing for the purchase of “valuable mineral deposits” extended to the purchase of water rights.16 While conceding that water is, in a sense, a mineral, the Supreme Court reasoned that “[t]he word ‘mineral’ is used in so many senses, dependent upon the context, that the ordinary definitions of the dictionary throw but little light upon its signification.”17 The word “mineral” is polysemous: it has a broad sense that could refer to any inorganic substance and a narrow sense that refers only to excavatable commodities like oil or metals. Given Congress’s long history of legislating on mining and water rights separately, the mining statute likely used “mineral” in the narrow sense that excluded water.18
Or take McBoyle v. United States,19 where the issue was whether flying a stolen plane counted as transporting a stolen “motor vehicle.”20 The term “motor vehicle” is polysemous: it can refer broadly to any conveyance or narrowly to conveyances moving on land.21 The Court held, in light of contextual beliefs about how the term was commonly used, that the narrow sense was operative.22 Similarly, in Taniguchi v. Kan Pacific Saipan,23 the question was whether translators of written documents were “interpreters.”24 The word “interpreters” is also polysemous: it can refer broadly to any translator or narrowly to oral translators.25 The Court held, again in light of contextual beliefs about common usage, that the narrow sense was operative.26 These are at least arguable examples of a statute’s context pragmatically enriching what the statute objectively asserts by fixing the operative sense of a polysemous word.
With these ideas in mind, let us return to textualism. Textualism, I propose, asks courts to enforce a statute’s objective asserted content.27 We should bear in mind, however, that textualism is not primarily about what courts should do when it is obvious what a statute means. If textualism were only about what courts should do in these “easy” cases, it would not be a controversial theory. Textualism primarily concerns what courts should do in cases where there is room to reasonably disagree over what a statute means. In such “hard” cases, textualism asks courts to enforce the most plausible interpretation of a statute’s objective asserted content, i.e., what a reasonable reader would have been most likely to infer that the legislature intended to assert.
That is, textualism limits the set of admissible arguments in hard cases: it confines judges to considering what a reasonable reader would have been most likely to infer that the legislature intended to assert rather than what the legislature “really” intended to assert or which reading of the statute best advances its purpose or maximizes social welfare. I recognize that textualists are not all of one mind about what their theory holds and, as Bostock illustrates, do not always agree on what it directs. I offer the foregoing as a constructive interpretation of textualism—one which aims to make textualism the best theory of its kind that it can be—while acknowledging that it may not capture everything that textualists say about their theory.
II. The “Compositional” Approach Is Not Textualism
Eskridge, Slocum, and Gries devote much of their Article to developing two cross-cutting distinctions between (i) compositional and empirical approaches to statutory interpretation and (ii) intensional and extensional approaches to the same.28 The authors argue that the Bostock majority employed a compositional, intensional approach to statutory interpretation, while the dissents employed an empirical, extensional one.29 In this Part and the next, I explain why neither the compositional nor the extensional approach is a version of textualism properly understood and why it is unlikely that any of the Justices in Bostock meant to apply the compositional or extensional approach.
According to the “compositional” approach, courts should treat a statute’s meaning as “the sum of the meanings of its parts and of the relations of the parts,” i.e., the sum of the conventional meanings of its words and its syntactic structure.30 In other words, the compositional approach asks courts to look to just the semantic content of a statute’s language. It is no different from what many textualists derisively call “literalism.”31 Literalism is not a theory that anyone claims to hold; it is a strawman of textualism that would have judges consider only the semantic content of a statute’s language, without regard to how the statute’s context might pragmatically enrich what it objectively asserts.32
Although the majority opinion in Bostock repudiated a compositional approach,33 parts of the opinion were suggestive of such an approach. The majority began by identifying the conventional meaning in 1964 of each word in the phrase “discriminate because of sex.”34 It assumed that “discriminate” meant “treating [an] individual worse than others who are similarly situated;” that “because of” referred to the standard of but-for causation; and that “sex” meant classification as male or female.35 Putting the pieces together, the majority held that Title VII prohibits an employer from taking an adverse action against an employee that it would not have taken but for the employee being male or female.36 An employer who fires an employee because the employee is gay or transgender clearly violates Title VII on this interpretation.37
From a textualist perspective, the majority started in the right place by considering the conventional meanings of Title VII’s words but then went astray insofar as it failed to adequately consider the statute’s context. Now, perhaps the majority saw no need to consider the statute’s context, since the dissents never gave an alternative interpretation on which the statute’s context made a difference (the dissents contented themselves with making the purely negative argument that, whatever Title VII meant, it could not have meant what the majority thought it meant).38 But insofar as Title VII’s context did pragmatically enrich its objective asserted content, the majority never took that pragmatic enrichment into account. In short, while the majority did not set out to apply a compositional approach, it may have mistakenly done so.
Unlike the compositional approach, the “empirical” approach would have courts look to “empirical public meaning,” which depends “on the actual views that the American public would have had about the ultimate interpretive question in” a given case.39 This concept of empirical public meaning remains obscure to me, but it may involve another mischaracterization of textualism. If what the empirical approach claims is that courts should answer any question about how a statute applies by imagining how the American public would have answered that question at the time of enactment, then the empirical approach is not textualism. After all, the American public might have answered that question based not on the statute’s objective asserted content but on any non-textualist reason, like what people generally believed the law ought to be.
Perhaps I am misunderstanding what the authors mean by “empirical public meaning.” If empirical public meaning just is objective asserted content, then the empirical approach looks more like textualism properly understood. Regardless, the key takeaway is that textualism is not concerned merely with semantic content, nor it is about how the American public would have answered specific questions of statutory interpretation at the time of enactment. Rather, textualism asks courts to identify the most plausible interpretation of a statute’s objective asserted content and then apply that content to the case at hand.
III. The “Extensional” Approach Is Not Textualism
Eskridge, Slocum, and Gries also draw a second distinction between intensional and extensional approaches to statutory interpretation.40 A word or phrase’s intension is its sense or meaning, while its extension is the set of things that it refers to.41 The intension of “planet” is something like “an astronomical body orbiting a star.” Its extension is the set of all such things—Mercury, Venus, Earth, etc. The “intensional” approach asks courts to look to the intension of any word or phrase in a statute at the time of enactment (e.g., what “discrimination because of sex” meant in 1964).42 The “extensional” approach tells courts to look instead to the anticipated extension of any word or phrase in a statute at the time of enactment (e.g., the actions that most people in 1964 anticipated would count as discrimination because of sex).43
The authors strenuously argue that we should not take the extensional approach seriously.44 Suppose that a statute enacted in 1964 prohibits bringing any “vehicle” into a park.45 No one at that time knew about Segways. Nor did anyone know about the “Otto Bock Super Four” (a heavy-duty, motorized wheelchair that the authors use as an example).46 These objects had not yet been invented, so no one could possibly have anticipated in 1964 that the extension of “vehicle” might one day include them. Yet surely that fact should not, by itself, preclude our imagined ordinance from applying to Segways or the Otto Bock Super Four today.
I agree that the extensional approach is indefensible,47 but no sophisticated textualist takes such an approach.48 To the contrary, textualists are generally quick to acknowledge that the framers of a statute can fail to anticipate how the statute will apply. Most famously, textualists criticize the Supreme Court’s decision in Church of the Holy Trinity v. United States for failing to enforce the surely unanticipated application of a statute to an English pastor.49 What matters from a textualist perspective is not a word or phrase’s anticipated extension; what matters, as we have seen, is what a reasonable reader would have inferred that the legislature intended to assert.50
To be fair, there is a complication in this vicinity concerning abstract, evaluative concepts, like “cruel and unusual punishments” or “equal protection of the laws.”51 It seems that people can disagree a bit over the criteria for applying these concepts and yet still be talking about the same thing; it seems that they can hold different “conceptions” of the same concept.52 Some originalists about constitutional interpretation—and presumably some textualists about statutory interpretation—think that, when a legal text refers to such a concept, courts should enforce the conception of the concept that was popularly held at the time of the text’s enactment.53 So if a judge is interpreting the Eighth Amendment, the judge should try to understand the concept of cruel and unusual punishment as it was popularly understood in 1791 when that amendment was ratified.
Still, none of this suggests an extensional approach. It is wrong to say that looking to the popularly held conception of an essentially contested concept requires holding forever fixed that concept’s extension. What it involves is identifying the specific criteria that people used to apply a concept at a certain time and then using those criteria to discern the concept’s extension today. When Justice Scalia says that judges must look to eighteenth-century society’s comprehension of cruel and unusual punishment,54 we should understand him as proposing that judges ought to enforce the conception of that concept that was widely held at that time. And when he notes that capital punishment was widely practiced back then,55 we should take him as advancing that fact as evidence of what the popularly held conception of cruel and unusual punishment was. At least, that is the most charitable understanding of his position.
What about the Bostock dissents? Did they mistakenly apply the extensional approach? Parts of Justice Alito’s dissent do seem to suggest such an approach.56 Moreover, neither dissent ever put forward its own interpretation of Title VII’s objective asserted content.57 (I assume that this was a strategic choice, one which the dissents made because there is no interpretation of Title VII’s objective asserted content that both supports their position and fits easily with the Court’s precedents).58 Charitably understood, however, the dissents did not rest their conclusion on Title VII’s anticipated extension at the time of enactment but rather on what they believed the most plausible interpretation of its objective asserted content to be.
This was most evident in Justice Kavanaugh’s dissent. He emphasized that, when it comes to phrases like “discriminate because of sex,” courts must look to “the phrase as a whole, not just the meaning of the words in the phrase” because the phrase as a whole “may have a more precise or confined meaning than the literal meaning of the individual words.”59 Using the ideas introduced above, we can reframe his point as follows: it is not enough to simply aggregate the conventional meanings of the individual words in Title VII according to the rules of syntax. Doing so only gives the semantic content of the statute’s language, which is not the object of textualist interpretation. One must also consider whether context—including beliefs about how certain phrases were commonly used in 1964—pragmatically enriched what the statute asserted.
If this is what the dissents were arguing, their claim was not that Title VII only prohibits conduct that most people in 1964 anticipated it would prohibit. Instead, their claim was that the most plausible interpretation of Title VII’s objective asserted content does not prohibit the conduct at issue in Bostock. They never said what they took Title VII’s objective asserted content to be, but perhaps they could have argued that the word “discrimination” is polysemous. The majority assumed that Title VII employed “discriminate” in the broad sense of any adverse differential treatment.60 But Title VII’s context may have fixed a narrower sense as operative—like differential treatment that disadvantages members of one sex (or race, color, etc.) relative to another.61 In that case, Title VII would not prohibit the conduct at issue in Bostock insofar as it does not disadvantage males relative to females or vice versa.62
Whether this is the most plausible interpretation of Title VII’s objective asserted content, and whether it can be squared with the Court’s precedents, are difficult questions that would take us too far afield of our main purpose here. But it is at least an available interpretation, and perhaps it is what the dissents implicitly relied upon. The main point is that the extensional approach is not textualism, and it would be uncharitable to assume that the Bostock dissents were applying such an approach. Any plausible version of textualism must be an intensional version (though we should bear in mind that textualists may look to a word or phrase’s anticipated extension as evidence of its intension).
IV. Textualism Is Dynamic in Some Ways, Not in Others
Having discussed at some length what textualism claims and what it does not claim, we are now in a better position to consider Eskridge, Slocum, and Gries’s objections to textualism based on dynamism. The authors present three “temporal complications” for any theory of statutory interpretation that aims to enforce a statute’s “original public meaning.”63 Those complications are: (1) “societal dynamism,” where a word or phrase’s extension has changed since the time of enactment; (2) “linguistic dynamism,” where a word or phrase’s intension has changed since the time of enactment; and (3) “normative dynamism,” where prevailing legal norms or public morality have changed since the time of enactment.64
Societal dynamism poses a problem only for the extensional approach. Consider again the 1964 statute prohibiting bringing any “vehicle” into a park. An extensional interpreter would assume that the extension of “vehicle” was frozen in 1964, such that the statute could not cover new types of vehicles like Segways (or perhaps even new token vehicles like a particular car manufactured after 1964!). But as we saw above, the extensional approach is not textualism. Textualism, properly understood, looks to a statute’s objective asserted content at the time of enactment and so allows that the extensions of the words or phrases that a statute uses can expand or contract over time, as new types of objects are introduced or removed from the world.65
Significantly, the authors misdiagnose the source of interpretive difficulty in their central examples of societal dynamism. They suggest that the reason why it is hard to tell whether a 1964 statute about vehicles extends to Segways or the Otto Bock Super Four has to do with those objects being invented after the statute was enacted.66 But these interpretive questions would be no easier if Segways or the Otto Bock Super Four preexisted the statute. The issue is not societal dynamism but vagueness. The word “vehicle” is vague: it admits of borderline cases, in which there simply is no answer as to whether a given object counts as a vehicle or not. Segways and the Otto Bock Super Four are (or at least might be) such borderline cases. While these examples may present a problem for textualism, the problem has nothing to do with dynamism.67
The authors also object that, if textualists agree that the extension of a term can change over time (which, of course, they do), then textualists need (a) “a normative theory of why that extension should not also change due to linguistic or normative dynamism” and (b) a “coherent methodology for distinguishing among the forms of dynamism.”68 I find part (a) of this objection baffling. The arguments that textualists use to justify their position—e.g., that it respects legislative compromise,69 constrains judicial reasoning,70 and promotes fair notice71—are precisely normative arguments against allowing courts to consider linguistic or normative dynamism. These are arguments that changes to the content of the law should come, insofar as possible, from legislative action and not from changes in the intensions of words or prevailing norms.72 What additional normative arguments would the authors have textualists make?
Part (b) is theoretically interesting but of little practical consequence. The authors contend that the border between societal dynamism (which textualists take into account) and linguistic or normative dynamism (which they do not) is fuzzy. That seems right, but these categories are still easily distinguished in most cases. That “vehicle” may now apply to Segways is an instance of societal dynamism; that “domestic violence” once referred to insurrections but now more commonly refers to physical force against a domestic partner is an instance of linguistic dynamism. The authors themselves seem to acknowledge that some cases are easily categorizable as one form of dynamism or another.73 I doubt that the practical usefulness of textualism requires anything more than this rough ability to distinguish between these categories.
In the end, the authors’ only criticism of (any plausible version of) textualism is that it does not permit judicial “updating” of statutes in response to normative or linguistic dynamism, i.e., in response to changes in the meanings of words or prevailing legal norms or public morality.74 But that is hardly news to textualists; after all, from their perspective, their theory’s incompatibility with such updating is a feature, not a bug. To adjudicate between textualists and Eskridgean or Posnerian interpreters who would have judges update statutes,75 we need moral or political arguments about the proper role of judges, and the authors offer little of the sort.
V. The Meaning of “Sex”
Justice Samuel Alito’s dissent in Bostock cited numerous dictionaries defining “sex” as classification as male or female around the time of Title VII’s enactment.76 But as Eskridge, Slocum, and Gries note,77 Justice Alito overlooked that those same dictionaries also defined “sex” as “[t]he sphere of behavior dominated by the relations between male and female” and “[p]henomena of sexual instincts and their manifestations.”78 What these alternative definitions suggest is that “sex” in 1964 was polysemous: it could refer narrowly to classification as male or female but also more broadly to all that is culturally associated with being male or female (similar to what we today might call “gender”). Simply piling on dictionary citations cannot tell us anything about which of these two related senses of “sex” was operative in Title VII.
That said, Title VII’s context obviously fixed the narrow sense of “sex” as operative. A reasonable hearer at that time would have known that a phrase like “discriminate because of sex,” especially in an antidiscrimination law, generally employs the narrow sense.79 Moreover, a reasonable hearer would have known that Congress was nowhere near progressive enough to assert a prohibition against discrimination because of anything culturally associated with being male or female. After all, many laws at the time did discriminate based on cultural stereotypes regarding whom members of each sex should be attracted to,80 and there was as of yet little public awareness that this was morally wrongful.81 Thus, while “sex” in 1964 could refer broadly to all that is culturally associated with being male or female, no reasonable hearer would have inferred that Congress intended to use the word that way in Title VII.
Eskridge, Slocum, and Gries rely on corpus linguistics to reinforce what dictionaries from around 1964 already show—namely, that “sex” in 1964 could refer to what we would call “sexual orientation” and “gender identity” today.82 Specifically, the authors detail the results of a search in the Corpus of Historical American English for uses of “sex” around the time of Title VII’s enactment and argue that those uses prove that “sex” could be used in the broad sense discussed above.83 That seems entirely correct. The problem is that the authors imply that, just because Congress could have used “sex” in this broad sense, textualists must assume that Congress did use “sex” in the broad sense.84 But that conclusion does not follow.
Indeed, while Eskridge, Slocum, and Gries find many instances of “sex” or its derivatives being used in the broad sense, all of these are instances of “sex” or its derivatives being used adjectivally—e.g., “sex roles,” “sex relations,” “sexual division of labour,” “sexual attitudes,” and “unusual sex practices”85 The authors apparently did not find a single instance of “sex” being used in the broad sense in a syntactic structure resembling “discriminate because of sex,” where “sex” is used as a noun. Thus, if anything, the authors’ evidence supports the opposite conclusion from the one that they reach: it suggests that a reasonable hearer in 1964 would have inferred that Congress did not intend to use “sex” in the broad sense because it was not common at the time to use “sex” in the broad sense in a phrase remotely like “discriminate because of sex.”
What Eskridge, Slocum, and Gries’s work shows is that it is semantically permissible to interpret “discrimination because of sex” to refer to discrimination based on sex stereotypes and thus to discrimination based on an individual’s sexual orientation or gender identity. From a non-textualist perspective, this is a significant conclusion. A purposivist or pragmatist could argue that courts ought to adopt this semantically permissible reading of the statute because it is better adapted to achieving Title VII’s purpose or maximizing social welfare. There is much to be said for such an argument, but a textualist cannot follow a purposivist or pragmatist down that road; a textualist cares not about whether an interpretation of a statute is semantically permissible but about whether it is the most plausible interpretation of the statute’s objective asserted content.
I am mindful that this part of the authors’ Article grew out of an amicus brief and has more of an advocate’s tone.86 Eskridge has previously advanced forceful arguments for the interpretation of Title VII that prevailed in Bostock,87 and all of the authors should be congratulated on their success in helping to bring about an equitable shift in the law. But an insubstantial argument made in support of a normatively good outcome is no less an insubstantial argument, and it is important to see that the authors’ corpus-linguistics evidence is, from a textualist perspective, utterly inconsequential. In this respect, the authors’ work illustrates a danger of using corpus linguistics as an aid to textualist interpretation: it is not enough to show that a legislature could have used a word some way; one must show that a reasonable hearer would have inferred that the legislature did use the word that way.
Conclusion
I cannot address all that the authors have to say in the short span of this Essay, and I fear that some of my criticism may seem too quick. What I hope to have shown is that textualism, charitably understood, does not look merely to the semantic content of a statute’s language, nor does it aim to enforce the anticipated extensions of words or phrases at the time of enactment. While it is dynamic in the sense that it permits words or phrases to pick out new sorts of objects, that does not put any pressure on textualists to also take into account linguistic or normative change post-enactment. Finally, textualists should be wary of corpus-linguistics arguments like the one that the authors advance: such arguments might be helpful, but only insofar as they account for the gap between how a word can be used and how a legislature used it on a particular occasion.