Introduction

Language plays a central role in all realms of human life, and not least in law (e.g., Travers and Manzo 2016). At the pinnacle of the American legal system is the US Supreme Court, which considers cases of national importance involving the interpretation of federal law and the constitutionality of laws at all levels. While it is the Court’s written decisions which are binding on lower courts, as well as on the Supreme Court itself inasmuch as it respects its own precedents, the road to a written decision involves a great deal of talk. One place that occurs is in “conference,” when the justices talk to one another behind closed doors (Black and Johnson 2018). Another, much more public, venue for talk is during oral arguments, when advocates present their cases and answer questions from the justices.

One characteristic activity during oral arguments is discussion about hypotheticals posed by the justices. “Today, an advocate must, more than ever before, prepare himself for a stream of hypothetical questions touching not only on his own case but on a variety of unrelated facts and situations” (Prettyman 1984, 555), wrote one Court observer four decades ago. In fact, hypotheticals are an important component of the Socratic method used to instruct law school students in the application of abstract legal categories to diverse scenarios (Mertz 2007), but the justices use hypotheticals for a distinct purpose: to explore the consequences of a particular decision for future cases that may arise, and that may end up in front of the Court. More tactically, a hypothetical may be used to challenge an advocate’s position, by identifying an undesirable consequence of a ruling in his or her side’s favor. When that is the case, it shades into a “slippery slope” argument, particularly in cases involving freedom of speech. “The warning is frequently heard that permitting one restriction on communication, a restriction not by itself troubling and perhaps even desirable, will increase the likelihood that other, increasingly invidious restrictions will follow” (Schauer 1985, 363). In Schauer’s (1985) terms, the slippery slope argument draws the Court’s attention away from the details of the “instant case”—the case ostensibly before it—to the imagined “danger case.” In so doing, hypotheticals may also provide grounds for ultimately untethering the law from the past through a decision that overturns precedent, contrary to the principle of stare decisis (Segall 2011).

While all Supreme Court watchers know that hypotheticals are a staple of oral arguments, there has been no research on how, exactly, they are constructed, advanced, pressed, defended, answered, and evaded. One reason is that social-scientific research on the US Supreme Court has been overwhelmingly quantitative. With respect to oral arguments, that means seeking correlations between counts of verbal behaviors (e.g., number of words uttered or questions asked) and pre-existing justice or advocate attributes (e.g., Jacobi and Schweers 2017), or between justice behaviors and eventual votes (e.g., Jacobi and Sag 2019). Another reason is that qualitative research on judges and judging, insofar as it exists (e.g., van Oorschot 2021), takes an ethnographic approach that does not prioritize the production of audio recordings, much less their careful analysis. Consider, for example, Latour’s (2010) study of France’s Council of State (Conseil d’Etat), its Supreme Court for administrative law. Though Latour insists on the fine-grained analysis of judicial practice, he did not record the deliberations of the judges he studied, relying instead on field notes to reconstruct their exchanges. While this no doubt facilitated his access to the Court’s backstage, it also set severe limits on just how fine-grained his analysis could be—like trying to do astronomy by making celestial observations with the naked eye and recording them by hand (Shapin 1994, 266–291). His conclusions, consequently, mainly consist of high-level characterizations of how judges operate, as they try to make the smallest possible adjustments to the law in order to meet novel circumstances. One thing that concerns them is how their decisions in the present may create new problems in the future, though the French Council seems less inclined than the US Supreme Court to use this as a justification for undoing precedent.

How, precisely, people do things with language, in an interactional setting, is the purview of conversation analysis (CA). A key insight of CA is that details matter. Those include exact word choice (Heritage 1984; Pomerantz 1984), the duration of pauses between speaking turns (Stivers et al. 2009), the timing of interruptions (Gibson 2005; Schegloff 2002), and the extent of overlapping talk (Schegloff 2000). Such minutiae are both telling and consequential. For instance, a hesitation may indicate that the recipient of a request is about to decline it, while providing the requester the chance to modify its terms or rescind it entirely (Sacks 1987). The only way to capture such details is through audio recording, and the only way to represent them in print is through technically precise transcription. Extemporaneous fieldnotes like Latour’s (2010) are not enough, nor are conventional transcripts, even when based on a recording (e.g., Scheffer 2010; see also Gibson 2022).

A particularly relevant example, because it involves talk about possible futures, is Gibson’s study (2011a, 2012) of the deliberations of President John F. Kennedy’s inner circle (the ExComm) during the Cuban missile crisis. The contribution of that work was not the discovery that the ExComm talked about the future, but the identification of the exact conversational resources and practices applied to this task. Those included procedures for collaborative storytelling that were different than those used in ordinary conversation to talk about the past (Lerner 1992), because they had to accommodate the open-endedness of the future and contestation about what that might bring (Gibson 2011b). Gibson also documents the strategic use of interruptions and overlapping talk to interdict, redirect, and suppress a story heading in an undesirable direction. This research would not have been possible without the release of the audio recordings secretly made by President Kennedy.

This article represents the first serious foray into the study of Supreme Court oral arguments using CA.Footnote 1 It focuses on a key exchange between three conservative justices and the government’s advocate in the March 2009 oral arguments for Citizens United v. Federal Elections Commission, one that, according to legal experts, changed the course of campaign finance regulation, or at least portended that radical change was on its way (Southworth 2024; Toobin 2013). This exchange involved the construction of a hypothetical involving an imagined ban on books that, the justices feared, or purported to fear, might follow from a Court ruling in the government’s favor. Cornered by three justices pursuing similar lines of questioning, and receiving no assistance from the more sympathetic justices, the advocate conceded that the government possessed the power to restrict (if not exactly ban) the publication of books, though he tried in various way to resist the terms of the hypothetical and the slippery slope it indicated.

The reason for focusing on a single episode is that, like an extended family therapy session (Bilmes 1985), or negotiation with a hostage-taker (Garcia 2017), or the act of coaching someone to keep a secret without saying what the secret is (Gibson 2016), talk about a hypothetical unfolds over several minutes, during which its consequences are progressively realized, neutralized, or some combination of the two. In other words, if we want to understand the talk engendered by a hypothetical, we have to follow that through for several minutes, and space limitations prevent me from considering more than one such episode here. Fortunately, decades of research support the supposition that there is “order at all points.” As Sacks put it, “[t]ap into whomsoever, wheresoever, and we get much the same things” (Sacks 1984, 22). From this perspective, analyzing a single episode of talk is less like surveying a single person than extracting an ice core pulled from a glacier or a pale of water lifted from a lake: in each case, what we discover therein provides us with at least an initial reading of the sort of things we could expect from additional samples taken from the same general location.

One contribution of this article is to our understanding of how temporality is navigated and wielded in legal argumentation, the connection between law and time—for instance, in the organization of evidence to represent the past in the expectation of a future trial (Scheffer 2004; van Oorschot 2018)—being an emerging area of study (Beynon-Jones and Grabham 2018; Scheffer 2010). This research also contributes to CA in three ways. First, and most straightforwardly, it extends the study of institutional talk (Drew and Heritage 1992b) to the Supreme Court (and appellate courts more generally); previous work on courts has been limited to trial courts (e.g., Drew 1992; Heritage and Clayman 2010). Second, it identifies some of the ways in which conversational rules and resources are deployed for rhetorical purposes, something that has received only modest attention within CA (e.g., Atkinson 1984; Drew 1992; Matoesian 1997; Pomerantz 1986). Third, it provides a model for the kinds of questions that can be asked, and the kinds of inferences that can be drawn, given information about what speakers can say, or are likely to want to say. This information, generally lacking in CA, is at least partly available in the present context, for we know about the relevant prior court rulings, the justices’ ideological predilections, and the arguments that advocates want to advance inasmuch as they were spelled out beforehand in written briefs.

I begin with some background on Supreme Court oral arguments, particularly how they are conducted and how argument behavior has changed over time, and touching on the question of whether oral arguments are consequential. I also briefly describe the recordings regularly made by the Court. Then I turn to Citizens United. After providing background about the case, I summarize the two sides’ written briefs, and the oral arguments preceding the focal exchange. Then follows the empirical analysis. In the discussion, I summarize the argument, consider its implications for our understanding of the Court’s power (including its power over temporal points of reference), and expand upon the article’s contributions to conversation analysis.

Supreme Court Oral Arguments

The Supreme Court has been hearing oral arguments from its very beginning, in 1791. At first, arguments for a case could last for days, though over time, written briefs have ascended in importance and the time allocated to “orals” has decreased enough that it is now possible for the Court to hear two cases per day. Oral arguments are held each year from October through April, three days a week in two week stretches with breaks in between.

For a conversation analyst, the first step to understanding talk in any institutional setting is grasping the operative “speech-exchange system”—the rules for allocating speaking turns and the types of content expected from those who receive them. In ordinary conversations, speaking turns are allocated through two principal mechanisms. First, if someone is addressed with the first part of an “adjacency pair,” such as a question, they are expected to speak next so as to provide the second part—in this case, an answer—or to make some excuse for not doing so. Second, if no one is addressed in this way, the first person to speak at a possible speaker transition point, such as at the end of a clause, gets the floor (Sacks et al. 1974). These rules mostly ensure that only one person speaks at a time without the need for top-down direction. They also afford participants wide latitude in terms of what they can say, so long as it is relevant to whatever was said last (and sometimes even this is not required [Maynard 1980]), though the range of options in the immediate aftermath of an adjacency pair first part is more restricted.

These rules are modified to meet the needs of particular institutional settings, such as in the classroom when students raise their hands and are called upon by the teacher. More pertinently, there has been important work on the speech-exchange system in one particular legal setting: when witnesses testify, and are examined, in court. Atkinson and Drew (1979) and Drew (1992) detail its distinctive features. Most of the associated talk is required to adhere to a specific format, namely a succession of question-answer adjacency pairs in which the examining lawyer asks a question and the examined witness answers it. However, some (e.g., badgering) questions are out of bounds, as are some answers (e.g., those relying on hearsay).Footnote 2 Lawyers can further limit the responses of witnesses by asking polar (i.e., yes/no) questions and insisting on a succinct answer, though the witness has no formal obligation to comply unless the judge demands it (Lubet 1992). That they are in charge of asking questions that witnesses are required to answer also gives lawyers power over the topic. In spite of the centrality of question-answer adjacency pairs to all of this, lawyers and witnesses may do other sorts of interactional work, under cover of asking questions and answering them, such as insinuating guilt and defending against such insinuations. Finally, that talk is mainly dyadic means that third parties have limited opportunities to speak: opposing counsel may object (e.g., to an inappropriate question or answer) and judges may instruct, but essentially no one else may talk, on pain of punishment. Because jury members have to keep quiet, the attorney cannot judge from their verbal responses whether they registered the relevance of a witness answer. Therefore, examining attorneys do various things to help the jury along, such as pausing for a long time after a witness answers to give the jury time to process the answer, and repeating all or part of it.

The Supreme Court speech-exchange system has some things in common with that of a trial court, but differs in important respects. During the era of Citizens United, arguments were limited to an hour, with each side speaking for thirty minutes. (As explained earlier, this had been longer, and a recent change in format increased the length again.) After being introduced by the chief justice, advocates are granted two minutes of uninterrupted time in which to make, or being making, their argument, though at the time of Citizens this was merely a norm, not yet codified, and it was not always respected. Then justices begin asking their questions, and readily interrupt advocates to do so. In this, they are empowered by the Guide for Counsel,Footnote 3 which instructs advocates to stop speaking when they are interrupted but also to never interrupt a justice.

In some ways, this resembles trial court, with justices acting like lawyers questioning (and interrupting) actual lawyers who are serving as witnesses to their clients’ positions. One difference, however, is that justices are more tolerant of circuitous responses to polar questions. A second is that while there is only one advocate at the podium at a time, there are usually nine justices, any of whom can interject, so that the system is not dyadic, though no one else has speaking privileges. A third, related to this, is that advocates are trying to convince justices who, unlike jurors, can verbally respond; however, most will not respond after any particular advocate remark or response, so that their understanding(s) will remain mostly hard to gauge. A fourth difference is that justices can use their speaking turns to make extended arguments of their own, though these are usually packaged as questions at the outset (“What do you say to the argument that…?”) or at the conclusion (“What’s your response to that?”). A fifth is that the advocate is the default speaker, permitted and expected to continue presenting his or her prepared remarks when not answering questions, which gives him or her considerable control over the topic (the precise legal issue being addressed), so long as the justices do not have other ideas. A sixth difference, separate from the speech-exchange system but too important not to mention, is that advocates aim to persuade the justices whom they address, so have every interest in being cooperative and professional, unlike a cross-examined witness who may become surly and evasive when confronted by a lawyer who makes no pretense of being undecided (e.g., Drew 1990, 1992; Matoesian 1997).

Additional advice is offered by the Guide to help advocates cope. For example, they are advised to prioritize their points so that the most important ones are not neglected in the event that a barrage of questions make it impossible to get through them all. Further, they are encouraged to start by responding to their opponent’s most compelling arguments, and also by providing better answers to questions their opponent struggled with. Finally, advocates are enjoined to “make every effort to answer the questions directly. If at all possible, say ‘yes’ or ‘no,’ and then expand your answer if you wish” (p. 11)(also Scalia and Garner 2008, 193-194).

This speech-exchange system obviously leaves justices with a great deal of latitude, and with it they have become especially assertive in recent years. Starting around 1995, Jacobi and Sag (2019) show, justices began talking more, specifically by making more assertions (while the number of questions remained about the same), and also interrupting advocates more, and being interrupted more by them in turn,Footnote 4 all of which Jacobi and Sag (2019) take to be a reflection of growing political polarization. They also report, as does Malphurs (2013, 104-105, 120-121), that justices have lately taken to speaking more often to the side that they ultimately vote against, suggesting they arrive to oral arguments ready to go on the attack. Thus, advocates can expect to be confronted by aggressive justices, though a justice (particularly, in his day, Antonin Scalia) will occasionally step in to lend a helping hand to the side he or she supports, such as by asking an easy question or suggesting a better response to an earlier one.

One motivation for the study of institutional talk, though often unstated, is that such talk matters—for what suspects say when they are questioned by police (David et al. 2017), for whether a child receives an autism diagnosis (Maynard and Turowetz 2022), for what information a patient shares with a doctor (Heritage et al. 2007), for what stories end up on a newspaper’s front page (Clayman and Reisner 1998), and, possibly, for what decisions heads of state make (Gibson 2012, 2018). An obvious question, then, is whether any of what happens during oral arguments, including discussions about hypotheticals, matters. Justices have occasionally affirmed the consequentiality of oral arguments. For instance, Justice Ruth Bader Ginsburg once wrote, “I have seen few victories snatched at oral argument from a total defeat the judges had anticipated on the basis of the briefs. But I have seen several potential winners become losers in whole or in part because of clarification elicited at oral argument” (qtd. in Malphurs 2013, 24). Johnson (2004) provides quantitative evidence that oral arguments are consequential, finding that justices ask questions about many issues not covered in briefs, issues that arise only during oral arguments are regularly discussed during conference (judging from memoranda and meeting notes), and such issues are regularly addressed in majority opinions. This last fact suggests that even in cases where no justice is persuaded to change his or her intended vote by oral arguments, the latter may well influence how the decision is written—in particular, whether it is written narrowly to apply only to the case at hand and those very similar to it, or broadly, with ramifications for a wide variety of future cases (Malphurs 2013, 23). There is also some limited quantitative evidence that justices are sometimes persuaded by an adroit oral argument to flip their opinion to align with it (Ringsmuth et al. 2012). In sum, there is reason to think that orals are consequential, though it is hard to say how often or what the exact magnitude of the effect is.

Recordings of oral arguments have been made since 1955, and those are available online,Footnote 5 as are transcripts dating back to 1969.Footnote 6 Importantly, because the US Supreme Court does not permit cameras, my analysis is limited to the audio recording. This is not ideal, as justices may, for instance, convey their displeasure through facial expression or hold the floor with a hand gesture (Malphurs 2013, 91). In the very least, when something in the recording evades “logocentric” (Ayaß 2015) explanation, we must consider the possibility that some nonverbal behavior (or other physical circumstance, like disorganized notes) contributed.

Citizens United v. Federal Elections Commission

Supreme Court oral arguments are saturated with context, including details of the case and the statute that is being challenged, applicable precedents, and written briefs. As a result, advocates operate within a legal framework within which they are pre-committed to making some arguments and steering clear of others. To the extent that that framework can be sketched in advance, we are in a better-than-usual position to answer the quintessential conversation-analytic question, “Why that now?” (Schegloff and Sacks 1973). Also relevant, for an analysis that picks up well into arguments, is everything that has been said up until that point; for instance, an advocate may wish to avoid repeating himself or herself, or may need to create an opportunity to complete a point that was left unfinished earlier (as in Bilmes 1985). I lay out each layer of context in turn.

Background

The history of campaign finance law is complicated (e.g., Urofsky 2020). Very briefly, the 1971 Federal Election Campaign Act (FECA) consolidated earlier laws about disclosures of campaign contributions (Pub. L. No. 92-225, 86 Stat. 3). In 1974, this was amended and greatly bolstered to include limits on contributions and independent expenditures (money spent on behalf of a campaign by someone other than the candidate); this also established the Federal Election Commission (FEC), charged with enforcement. In 1976, the Act was amended again, after parts were struck down in Buckley v. Valeo (424 US 1 [1976]). Further complications ensued, but what is important for our purposes is that by the turn of the twenty-first century, the FEC had the power—though it was frequently contested—to restrict expenditures by corporations on behalf of a candidate when that involved “express advocacy” (e.g., in a television advertisement) that included one of the “magic word” imperatives such as “vote for X” (Mayer 2009). (Buckley distinguished this from “issue advocacy,” which it declared exempt from regulation even when it identified a candidate’s position on the issue.) In particular, corporations were banned from paying for such messages using general treasury funds, and required instead to use segregated money raised for the purpose (as part of a PAC, or political action committee) from individuals associated with the corporation.

In 2002, President George W. Bush signed the Bipartisan Campaign Reform Act (BCRA), sometimes referred to as McCain-Feingold, after its original sponsors (Pub. L. No. 107-155, 116 Stat. 81). As a further amendment to FECA, this banned any corporation or union from using its general treasury funds to finance “electioneering communication,” which it defined as “any broadcast, cable, or satellite communication which refers to a clearly identified candidate for Federal office” within sixty days of a general election or thirty days of a primary (§201), with exceptions for traditional journalism and also certain nonprofit corporations on the condition that they did not accept contributions from for-profit entities.

BCRA was mostly upheld when challenged in McConnell v. Federal Election Commission the following year (540 US 93 [2003]). In 2007, in Federal Election Commission v. Wisconsin Right to Life, Inc., however, the Court declared (in an opinion authored by Chief Justice John Roberts) that BCRA was unconstitutional when applied to advertisements that fell short of express advocacy. The ruling also devised a standard by which to determine whether an ad that fell short of saying “vote for X” was nonetheless partisan enough to be the “functional equivalent” of express advocacy. In its McConnell ruling, the Court had said that this was a matter of the intent behind the ad, but with Wisconsin Right to Life the Court aimed for something marginally less indeterminate: “an ad is the functional equivalent of express advocacy only if the ad is susceptible of no reasonable interpretation other than as an appeal to vote for or against a specific candidate” (551 US 449, 469-470 [2007]).

In 2007, the nonprofit corporation Citizens United sought to use a video-on-demand service to disseminate a documentary critical of Democratic presidential candidate Hillary Clinton, entitled Hillary: The Movie. This ran afoul of BCRA because of its partisan content, timing (soon before the onset of Democratic primaries), and funding (general treasury funds to which some for-profit corporations had contributed). Anticipating FEC action against it, Citizens United sought declaratory and injunctive relief from the federal district court in Washington, D.C., which ruled in the Commission’s favor (530 F. Supp. 2d 274 [DC 2008]). The court also upheld BCRA reporting, disclosure, and disclaimer requirements as they pertained to advertisements for the documentary. Citizens United appealed to the Supreme Court, which agreed to hear the case.

Oral arguments were heard in March of 2009.Footnote 7 Initially, Citizens United asked only for relief from BCRA “as applied” to its particular circumstances—a nonprofit corporation providing an informative documentary to willing viewers, mostly paid for through individual donations. However, conservatives on the Court were eager to consider the “facial” case regarding the very constitutionality of BCRA, and ordered the case reargued in September so that this more sweeping issue might be addressed. The Court ultimately ruled in favor of Citizens United (558 US 310 [2010]), holding that limitations on corporate campaign expenditures are an unconstitutional infringement on free speech. This meant nullifying part of BCRA and reversing earlier Court decisions: Austin v. Michigan Chamber of Commerce (494 US 652 [1990]), which affirmed the government’s power to limit corporate campaign expenditures, and the part of McConnell that rested on that. The majority opinion went to great lengths to justify the facial ruling given the as-applied nature of the original challenge, something that violated its oft-stated (but oft-disregarded) preference for narrow rulings (Fallon 2011). The Court had also circumvented Court rule 14.1(a) against considering questions beyond those identified in the original petitionFootnote 8 through the unusual device of ordering the case reargued (Jackson 2020) and dictating what those questions would be.

In this article I focus on the original (March 2009) arguments. These were preceded by written briefs, which I now summarize, focusing on the arguments about the documentary itself and putting aside questions related to the advertisements for it. The briefs are important because they provided the material for the advocates’ planned remarks, and set limits on their responses to questions, for an advocate should never contradict his or her side’s written brief. That is, they provided the advocates with conversational parameters, giving us unusual insight into what the advocates could say when talking.

Written Briefs

Petitioner’s Brief (for Citizens United)

The petitioner’s briefFootnote 9 argued as follows: The first amendment says that “Congress shall make no law…abridging the freedom of speech” (1). As the Court asserted in FEC v. Wisconsin Right to Life (hereafter WRTL), that means that a court should “ensure that a compelling interest supports each application of a statute restricting speech” (2). Unlike the short commercials that BCRA was meant to address, Hillary was directed at people who wanted to watch it and whose minds were unlikely to be changed by it. Also, funding for the documentary came overwhelmingly from individual contributions to an overtly ideological organization, not from donations by for-profit corporations whose shareholders had no say in how their money was being spent (though a small amount of money was obtained from such corporations). The documentary is not express advocacy, or the functional equivalent, but rather about Hillary Clinton’s public record. There was no risk of corruption because none of her Democratic rivals (in the primary) would have considered it a contribution in need of repayment. In addition, it was permissible to release the documentary in theaters and on DVD, so it was arbitrary to target the video-on-demand release.

This was mainly an as-applied argument, but the brief also took a facial swipe at Austin, which, again, affirmed (pre-BCRA) the government’s compelling interest in regulating corporate political speech. This, the brief said, “was wrongly decided and should be overruled” (13, 30).

Respondent’s Brief (for the government)

The government’s briefFootnote 10 argued as follows: The Court has recognized that Congress has a compelling interest in ensuring election integrity and in regulating corporate and union donations—corporations because they might advocate for positions without the consent of their shareholders. Hillary qualifies as the functional equivalent of express advocacy by the criterion set forth in WRTL because it focuses on Clinton and her qualifications for office. Its duration does not matter, nor does the fact that people would have had to voluntarily opt to watch it—it still could have been consequential had it motivated the base, and it might have swayed some undecided swing voters. That BCRA targets televised advertisements is just because they were seen as most pernicious at the time that statute was written; other long-standing restrictions cover other modes of communication. Moreover, Citizens United used corporate treasury funds, which BCRA forbids. In fact, we do not really know if most funding came from individuals; even if it did, their views might not have been represented by the documentary; and regardless, the FEC needs a clear rule (like “no contributions from for-profit corporations”) for the sake of administrability.

Petitioner’s Reply

The petitioner respondedFootnote 11 as follows: The government failed to make the case that this particular documentary posed any risk of corruption. Moreover, “[t]he government’s position is so far-reaching that it would logically extend to corporate or union use of a microphone, printing press, or the Internet to express opinions—or articulate facts—pertinent to a presidential candidate’s fitness for office” (2); it’s only by “Congress’s grace” (6) that BCRA does not extend to these. Opt-in communication is not as influential as a traditional television advertisement and thus there is less space for corruption. Most donors were individuals and that needs to be considered before speech is stifled. Administrability isn’t a “compelling interest.” The documentary isn’t the functional equivalent of express advocacy because it can be interpreted in another way, as “a documentary designed to educate the public” (18).

Oral Arguments

The advocate for the petitioner was Theodore Olson, famed conservative lawyer, former Solicitor General, and Court regular; this was his fifty-second appearance at it.Footnote 12 “Participation in the political process is the First Amendment’s most fundamental guarantee,” he began, “yet that freedom is being smothered by one of the most complicated, expensive, and incomprehensible regulatory regimes ever invented by the administrative state.” Under BCRA, he said, it was a felony for nonprofit Citizens United to air a documentary about a candidate for office though it could be “shown in theaters, sold on DVDs, transmitted for downloading on the Internet, and its message may be distributed in the form of a book.” It was the government’s burden, he continued, to establish that “each application of a restriction on that form of speech is a narrowly tailored response to a compelling governmental interest”Footnote 13 and it had failed to demonstrate that its interest in combatting corruption, or the appearance of corruption, required this particular act of censorship. “Indeed,” Olson said, “this documentary is the very definition of robust, uninhibited debate about a subject of intense political interest that the First Amendment is there to guarantee.”

At this point, at exactly the two-minute mark, the justices began their questions. Though sometimes confusing—and sometimes Olson seemed confused—these mostly pertained to two issues. The first was whether Hillary: The Movie qualified as issue, rather than candidate, advocacy. Olson claimed that it did but also noted that the decision in WRTL said that the distinction “dissolves upon practical application.” Justices David Souter, Stephen Breyer, and Ginsburg countered that the documentary seemed to expressly advocate for a candidate. Olson replied that, by that logic, any communication that expresses a “point of view” could be suppressed (a weak response that Breyer immediate challenged), and that the government has to demonstrate both that it has a compelling interest for a restriction on such speech and that the restriction is narrowly tailored so as to advance exactly that interest.

The second issue was the distinction between a ninety-minute documentary and a thirty-second advertisement. Olson was asked what made the latter subject to BCRA but not the former. Olson responded that the documentary was educational, and also that it was with short advertisements that Congress was demonstrably concerned when crafting BCRA, as “the things that were most potentially corruptive,” and that people “have no choice about seeing.”

Next it was the turn of the government’s advocate, Deputy Solicitor General Malcolm Stewart. WRTL, he said, defined “functional equivalent of express advocacy” and length was irrelevant to that definition. In this, and his side’s merit brief, we see Stewart’s preferred realm of argumentation, pertaining to the instant case—the applicability of BCRA, as interpreted through WRTL, to the details surrounding the planned release of Hillary. At that point, however, a mere thirty-eight seconds into Stewart’s argument, the justices commenced their questioning, which eventually fixated on the danger case of a book ban. Roberts started more whimsically, however: after wondering how anyone could know Congress’s intent in writing BCRA, he asked whether an advertisement for “candidate action figures” (which it would presumably name) would count as electioneering communication, and thus be subject to BCRA.

Analysis

“The oral arguments were about to take the case—and the law—in an entirely new direction,” writes Toobin (2012, 36) of this exchange, which lasted about four minutes. I divide this into four segments, or excerpts, consisting of Justice Samuel Alito’s first questions to Stewart and the latter’s eventual response; the remaining exchange between the two; Stewart’s exchange with Justice Anthony Kennedy; and his exchange with Chief Justice Roberts. To anticipate, the three conservative justices constructed a hypothetical (danger case) in which the FEC might undertake to prevent the publication of a book containing express advocacy for a candidate, though Hillary was not a book, though BCRA did not pertain to books, and though older campaign finance rules not ostensibly at issue in this case already empowered the government to do exactly that. In so doing, they made frequent reference to the Constitution, and sometimes more explicitly the First Amendment. Temporality was central to this, as captured in Fig. 1. Alito and Roberts, in particular, used the hypothetical to untether the discussion from precedent (the recent past), as well as from the details of Citizens (the instant case), in order to focus on an imagined future as seen through the lens of the most relevant distant past—the Constitution. This future, it was implied, would follow from the whole way of thinking embodied in the campaign finance regulatory regime that Stewart represented. In this way, the three justices set the stage for converting an as-applied challenge to BCRA into a facial one that went beyond it. Stewart tried to resist by anchoring his responses in precedent (especially in challenging the word “ban”) and reminding the Court that any restriction on books stemmed not from BCRA but from FECA. However, he was outnumbered by justices who, thanks to the local speech-exchange system, were entitled to interrupt at will and to dictate “what we’ve been talking about.”

Fig. 1.
figure 1

Timeline and summary of the argument

Full size image

Underlying this were several interactional practices, in addition to interruptions. Everything the justices said was a question, or supportive of a question, as evidenced by the fact that eventually Stewart had to be given the chance to respond. Justice Alito asked three straightforward interrogative questions, one beginning with “do,” a second with “what,” and a third with “would.” The first and third were polar, designed to favor a “yes” or “no” answer, or something close to it. Another polar question, from Chief Justice Roberts, was a tag question appended to a statement: “correct?” All of the others were statements presented to Stewart for affirmation or denial; these are known as declarative questions, and are also considered to be polar questions (e.g., Stivers 2010). An important subtype of declarative question involves paraphrasing someone’s foregoing talk with the goal of having that person affirm or reject that synopsis as adequate. Heritage and Watson (1979, 1980) call this a formulation. In Garfinkel and Sacks’s (1970) original coinage, this referred, rather amorphously, to a speaker’s characterization of “what we are doing,” and a better term for this particular variety might be reflected formulation. Most of the declarative questions produced by the justices were of this variety, amounting to conjectures about the advocate’s view of what the government could do in the hypothetical situation, based on what Stewart had said up to that point as well as his side’s written brief.

The “yes” or “no” answer (in response to a polar question) recommended by the Guide for Council is called type-conforming by conversation analysts. Less direct responses are type-nonconforming, and the more circuitous the response is—that is, the longer it takes for the speaker to actually communicate something equivalent to a “yes” or a “no”—the more nonconforming it is. Type-conforming responses are “preferred,” not just during oral arguments, per the Guide, but also in ordinary conversation. That means that the recipient of a polar question is expected to produce a type-conforming response, and will feel pressure to do so, or at least to make excuses for failing to do so (Raymond 2003).

These aligned pressures notwithstanding, almost all of Stewart’s responses were nonconforming. This was true when a simple response was readily available to him, but also when a justice’s reflected formulation had the property of being (to coin a term) multiply problematic, containing more than one objectionable supposition. This made it important for Stewart to identify what portion he was objecting to, and to create the space for himself in which to do this, something that a turn-initial type-conforming response might not have done had a justice immediately asked a follow-up question. The real danger of these multiply problematic questions, however, was that each forced Stewart to decide what to contest and what to leave unchallenged, as he could not expect to speak to more than one point, or to provide one discernible answer, before a justice interrupted with another question. At the level of conversational machinery, then, the three conservative justices constructed a hypothetical, and prompted Stewart to respond to it by presenting him with reflected formulations, some of which were multiply problematic. Stewart responded with type-nonconforming responses which bought him enough time to offer a precise, substantive challenge, but only one each time, and even then, the justices minimized the import of his responses, sometimes by simply ignoring them.

Figure 2 starts with Stewart’s response to Roberts’s question about advertisements for candidate action figures. In another context, this question might have been taken as facetious, but an advocate’s assumption must be that when a justice asks a question it has some serious import, whatever its actual intent, whatever its form, however it is delivered, and whatever response (e.g., laughter) it garners from others in the room. This particular question poses a hypothetical, shifting the discussion from the details of Citizens to an imagined advertisement, possibly with the goal of making BCRA look ridiculous, except that Stewart has an effective response: such an advertisement would indeed be covered by BCRA if it ran soon before an election (“at the right time”), except that, under WRTL, it would be unconstitutional to obstruct such an ad. The reason is not fully spelled out before Alito interrupts, but the phrase “other than” is presumably part of the WRTL formula for judging whether a message counts as the functional equivalent of express advocacy, of which an advertisement for candidate action figures would fall well short. That we can guess what Stewart was about to say means that everyone else could as well, including Alito. If we assume he was eager to interrupt, this could explain why he does so at this instant, just at the point at which Stewart’s point is discernible even if it remains incompletely voiced (Jefferson 1984).

Fig. 2
figure 2

Starting at 28:49

Full size image

Alito launches his own hypothetical in lines 8-21, though he does so in stages, by means of a succession of three questions that are only possible because Stewart does not immediately respond to the first, or to the second, contrary to one of the principal rules of turn-taking (Sacks et al. 1974). Granted, sometimes a speaker will rephrase a question if the recipient misunderstands it, or seems uncertain how to answer (Schegloff et al. 1977), but there are no signs of trouble from Stewart apart from lack of immediate response and, aside from the fact that he keeps revising his question, no indication from Alito that he considered either of the earlier versions of his question deficient (e.g., in the form of a self-repair token like “I mean”). Because this is the moment that the decisive hypothetical emerges, it is important to analyze Alito’s turn in detail, though this means speculating about Stewart’s reasons for not responding sooner, silence being harder to interpret than the spoken word.

Alito first asks, “Do you think the Constitution required Congress to draw the line where it did, limiting this to broadcast, cable, and so forth?” (ll. 8-11).Footnote 14 While foreshadowing Alito’s eventual slippery slope argument, on its own the question is oblique, as it is not obvious what is on the other side of “the line,” and this might explain why Stewart hesitates in responding—for eight-tenths of a second (which includes the 0.4-second inhalation) is a relatively long delay by the standard of English question-answer sequences (Stivers et al. 2009). Another explanation is that Stewart does not want to answer the question as posed, with a type-conforming “yes” or “no”. Were he to do so (per the Guide’s advice to “answer the questions directly”), the answer would need to be “no,” as that is the government’s position (judging from the brief and also from his eventual response), and that would be to walk right into Alito’s trap, of claiming for Congress even wider-ranging powers to regulate political speech. And because Alito could then immediately interrupt, Stewart might be denied the opportunity to justify his response or add caveats. Thus, the delay may be evidence that Stewart is about to offer a type-nonconforming response, one that resists the terms of the question, for such responses are typically delayed (Raymond 2003; Stivers et al. 2009).

Whatever the reason for Stewart’s hesitation, it allows Alito to ask a second question: “What’s your answer to Mr. Olson’s point that there isn’t any constitutional difference between the distribution of this movie on video demand and providing access on the Internet, providing DVDs either through a commercial service or maybe in a public library, providing the same thing in a book?” (ll. 11-19). Here, under cover of asking a question, Alito presents Stewart with a position, which he attributes to Olson: so far as the Constitution is concerned, one communication medium is equivalent to another. However, this actually misrepresents Olson’s position, if Alito is referring to the latter’s oral argument. Olson did not say that there is no “constitutional difference” between those media, but that as Hillary was available on DVD, on the Internet, and in theaters, and could be turned into a book, it was arbitrary for the FEC to single out the video-on-demand format. For Olson the implication of this was that the FEC should have refrained from interfering with the documentary’s video-on-demand dissemination. Yet, if we interpret Alito’s question in combination with those that preceded and followed it, he seems to be using the “no constitutional difference” argument, attributed to Olson, to advance the slippery slope argument about the possible scope of government regulation, that it might someday extend to books in libraries—a prognostication Olson had not made when he spoke, though the petitioner’s reply brief had warned of the threat to the “printing press” (2). In other words, Alito uses the device of reported speech (Holt 2009), with which he takes liberties, to repurpose a component of Olson’s as-applied argument (about the application of BCRA to Hillary) to advance his budding facial argument (about the unconstitutionality of campaign expenditure regulations generally).

From Alito’s perspective, the second question has a number of advantages. First, by omitting the reference to Congress and BCRA, it denies Stewart a response he will eventually employ, that at the time BCRA was passed, campaign finance law already regulated other media. Second, in attributing the view to Olson, the revised question makes Alito seem less partisan (as in Clayman 1988). A disadvantage, however, is that the second question is not a polar question and thus affords Stewart more latitude in responding.

This is followed by a 0.4-second pause. That Stewart does not respond this time may be due to the short time he is given to do so, though other explanations are available. One is that it is not clear, on grammatical grounds, whether Alito has finished his question, as there is no conjunction (in particular, “or”) before the terminal item on his list of ways the documentary might be provided (“providing the same thing in a book”). Perhaps, mindful of the injunction not to interrupt a justice, Stewart is waiting for him to continue.

Alito’s final question (ll. 19-21) piggybacks on the second one: “Would the Constitution permit the restriction of all of those as well?” With this, particularly given the future-conditional “would,” it becomes clear that the line of questioning is not about BCRA or Citizens United, but about the kind of laws that Congress might in the future enact on the basis of the constitutional interpretation that made BCRA possible. In this, Alito is walking the path laid by the petitioner’s reply brief: “The government’s position is so far-reaching that it would logically extend to corporate or union use of a microphone, printing press, or the Internet” (2). Note the change in tense at each step of Alito’s questioning, from past (“required”) in the first question, to present (“isn’t”) in the second, to future-conditional (“would”) in the third. In addition to revealing his real concern, an advantage of the third question over the second question is that the polar format is more restrictive than “what’s your answer to Mr. Olson’s point?” and thus more likely to extract from Stewart a reply that serves Alito’s purposes.

Now Stewart finally responds, and fairly quickly. His response is type-nonconforming. “I think the constant, Constitution would have permitted Congress to apply the electioneering communication restrictions,” he says, “to the extent that they’re otherwise constitutional under Wisconsin Right to Life, those could have been applied to additional media as well” (ll. 22-27). This does not merely resist the polar nature of the question, however, for Stewart’s response is about Congress’s past behavior, and thus is most easily heard as a reply to Alito’s first question, about what Congress did or did not have the constitutional authority to do when it crafted BCRA, except for the reference to WRTL, which was decided five years after BCRA was passed. Responding to the first question—now that it is apparent what lies on the other side of “the line”—permits Stewart to answer one of Alito’s questions, about events in the past, while evading, for the moment, the slippery slope implications of the final one (cf. Sacks 1987).

In lines 27-32, Stewart makes the first of two attempts to invoke the Federal Election Campaign Act, which BCRA amended. Judging from the later one (ll. 101-104), it is likely that Stewart is trying to point out that campaign finance restrictions predating BCRA already regulated printed materials expressly enjoining the reader to vote for a particular candidate. But this is a complex sentence, with an embedded clause (“which have been…”) postponing whatever Stewart meant to say about the FECA’s preexisting restrictions. Alito interrupts before Stewart can get to that—and quite possibly because the justice realizes that talk about FECA risks undercutting the incredulity he is affecting about BCRA, which is what this as-applied case had put squarely in the Court’s sites.

“That’s pretty incredible,” Alito begins (ll. 33 and 35 of Fig. 3). “You think that if a book was published, a campaign biography that was the functional equivalent of express advocacy, that could be banned?” Here we have the first reflected formulation, paraphrasing Stewart’s answer in order to isolate its “gist” (Heritage and Watson 1979). This is useful to Alito because it permits him to indicate the non-contiguous (and thus potentially ambiguous) referent of “that’s pretty incredible,” but it does additional work as well. For one thing, is enables him to smuggle in the charged word “ban” under the pretext of offering a paraphrase. In fact, Stewart had not used this word, though it did appear in both sides’ written briefs. For another, it allows Alito to insist on the future-conditional tense (“could be”) that Stewart had just avoided by responding to the first of the three initial questions. Alito also cleverly incorporates Stewart’s WRTL reference through formulaic mention of “the functional equivalent of express advocacy” (l. 37) which that ruling defined, which makes it seem as if he is paying attention to the deputy solicitor general’s answers even as he ignores the reference to FECA.

Fig. 3
figure 3

Starting at 30:05

Full size image

Stewart responds type-nonconformingly by resisting “banned,” explaining that Congress could “prohibit” a corporation from using its corporate (general) treasury funds, and instead require it to use PAC money raised specifically for the purpose (ll. 39-42, 44), as permitted by the Court in Austin. But note the tense concession: in speaking of what Congress “could” do, Stewart has yielded to Alito’s demand that he speculate about a possible future.

Alito again interjects (though Stewart completes his sentence in overlap): “Well, most publishers are corporations” (ll. 43, 45). This arguably uses Stewart’s reference to corporations in line 41 as an opening to introduce publishers (that are corporations), whose right to publish a book will seem constitutionally sacrosanct to any justice attempting to channel the Constitution’s framers. Alito also switches, in line 47, to the word “prohibited,” possibly echoing Stewart’s use of the word in line 40, but omitting the specifics about general treasury funding that Stewart had attached to it: “a publisher that is a corporation could be prohibited from selling a book?” he asks. Stewart could, in response, challenge this, just as he did “banned,” for oversimplifying, but he has another answer, that BCRA makes exceptions for the media. “Well, of course,” he begins, the “well” anticipating a type-nonconforming response (Heritage 2015), and “of course” indicating that what Stewart is about to say is something already known to Alito, or something that he should know (Stivers 2011). The two pauses in line 48 make Stewart sound slightly surprised at having to explain this.

Alito, however, brushes aside this response (l. 50), a clear demonstration of his indifference to the particulars of the reigning regulatory system and the instant case. “The government’s position,” he continues, “is that the First Amendment allows the banning of a book if it’s published by a corporation” (ll. 51-53), which calls on Stewart to reconcile the imagined book ban with the Constitution’s protection of free speech. Note that the reflected formulation is now of the “government’s position” specifically, a shift from Alito’s earlier questions (ll. 8, 12, and 35) that forces Stewart to represent the official position of his employer (see Goffman 1981, 144-146). Back is the word “ban” (in the form of “banning”) even as Alito has abandoned the reference to publishers.

At this point, Stewart could again challenge the word “ban,” as he did in lines 39-42, though this might have been seen as obstinate, as if he did not understand that Alito considered his earlier response insufficient. Instead, Stewarts yields to the demand for direct constitutional exegesis, but he manages this craftily. The institutional press, he says, may have a “greater First Amendment right,” because the First Amendment mentions it specifically, but then he distances himself, and the government, from this position twice over, referring to this as a “potential” argument and then adding that this “question is obviously not presented here.” In this, he takes advantage of the fact that in lines 45-46, Alito noted that a corporation could be a publisher, to make a point that might have been difficult had he not. Also, Stewart’s last words may be intended as a reminder of Rule 14.1(a): “Only the questions set out in the petition, or fairly included therein, will be considered by the Court.”

Stewart then attempts to move on to something else when Justice Kennedy interrupts. Throughout Fig. 4, Kennedy takes a different tack, grounding his questioning in the very details of BCRA, and the instant case, that Alito tried to bypass, while simultaneously invoking the Constitution and imagining a hypothetical ban on books—basically touching down in each era of Fig. 1. His opening query in lines 61-66 replaces a hypothetical corporation with a less hypothetical (because Citizens United-like) “advocacy organization,” and refers to the thirty- or sixty-day period during which electioneering communications were restricted under BCRA. It also replaces the “government’s position” (to which Alito had just escalated) with the more ambiguous “your position,” and reverts to the word “prohibited.” It maintains Alito’s focus on the hypothetical book, however, asking whether a book produced by such an organization would be prohibited in the months leading to an election “under the Constitution.”

Fig. 4
figure 4

Starting at 31:02

Full size image

Kennedy’s query, ending in a reflected formulation, is multiply problematic, containing more than one objectionable supposition. This puts Stewart in the position of having to decide which to respond to given that he is unlikely to be able to respond to all of them before being interrupted again. First, Stewart can simply reject Kennedy’s formulation of the government’s position on the grounds that the thirty-/sixty-day window, specific to BCRA, only pertained to broadcast, cable, and satellite transmissions. Second, he can challenge the word “prohibited” in the same way that he earlier challenged “banned.” Third, he can insist on the condition that the imagined book would actually need to advocate for a particular candidate, as required by the decision in WRTL.

Stewart opts for the third approach, invoking Roberts’s definition of the functional equivalence of express advocacy in that decision (ll. 67-70), beginning not with a type-conforming “yes,” but rather with the conditions under which this would be so. But Kennedy interrupts as Stewart is in the midst of this, tweaking his hypothetical to bring it more into line with BCRA, so that it features a satellite communication that is a (Kindle) book. “The existing statute would probably prohibit that under your view,” he says (ll. 74-75). This is another multiply problematic (declarative) question. In response, Stewart can press his point about express advocacy, or he can object to Kennedy’s use of “prohibit,” though doing one of these would almost certainly come at the expense of doing the other. Moreover, neither of those responses would address the new element in Kennedy’s question: the Kindle book. This puts Stewart on the spot, for there is no denying that BCRA would govern the publication of such a book.

Accordingly, Stewart concedes BCRA’s applicability to such a book, but circuitously, with another type-nonconforming answer (ll. 76-79). He basically agrees that BCRA would apply to a Kindle book, but not in so many words, reviewing—and sounding almost apologetic—the types of communications covered by that statute, as if to say that any coverage of an electronic book would be incidental to the Act’s original purpose. Stewart also takes cover behind McConnell, which upheld most of BCRA, implicating “the Court” as partly responsible for this state of affairs (though Kennedy dissented in that case).

Kennedy interrupts Stewart right after the reference to McConnell, and before we learn exactly what use Stewart was going to make of that ruling, and, in lines 80-87, basically repeats his formulation, “just to make it clear”—a reasonable justification given Stewart’s circuitous response to the justice’s previous attempt. This time, however, Kennedy specifically asks about the “government’s position,” as Alito did in line 51, as if both ultimately thought it important to get Stewart to expressly speak on behalf of the government but neither thought to do that at the start. He also specifies “campaign advocacy” (ll. 83-84), which is the condition Stewart insisted on in lines 67-70. In lines 80-85, Kennedy’s delivery takes on a sing-song (and somewhat condescending) quality, imperfectly captured by the intonational arrows. At the end of his turn, he makes a sharper distinction than he had previously between what the Constitution permits and what the statute does (“it can be prohibited under the Constitution and perhaps under this statute”).

Now Stewart challenges the word “prohibited,” repeating the point he made in lines 39-42 apropos the word “banned” in response to Alito—though, finding that Kennedy has appropriated “prohibit,” Stewart opts for yet another synonym: “barred.” Essentially, in posing the same declarative question a second time, Kennedy has allowed Stewart the chance to provide a second answer: in lines 76-79, the answer was that a Kindle book would be covered, and now (ll. 88-92), it is that the corporation would not actually be “prohibited” from publishing such a book, merely that it would have to do so using PAC money.

At this point, Chief Justice Roberts takes over the interrogation (Fig. 5). “If it has one name, one use of the candidate’s name, that would be covered, correct?” he asks (ll. 95-96), continuing the string of hypotheticals and the focus on the imagined future. The “it” here is ambiguous, however. Normally this would refer to something already under discussion. However, that would be Kennedy’s hypothetical Kindle book, which Kennedy stipulated consisted of “campaign advocacy,” which would presumably not be limited to one use of a candidate’s name. Instead, Roberts seems to have something new in mind. Now, for the first time, Stewart agrees outright, but must have intended to add the requirement that the unspecified “it” involve at least a modicum of express advocacy, per WRTL. Regardless, Roberts is indifferent to the concession, talking over Stewart to spell out his hypothetical in more detail. This (which in fact does depart from Kennedy’s) involves a long book with a single line of express advocacy: “Vote for X.” “The government could ban that,” he says (l. 100).

Fig. 5
figure 5

Starting at 32:12

Full size image

As he had in lines 48-49 and 67-70 with respect to “prohibit,” Stewart lets “ban” go unchallenged in order to respond that such a message would indeed qualify as straightforward express advocacy, and as such would be regulated under FECA (ll. 101-104), with his turn-initial “well” again announcing that the response will be type-nonconforming. But then Roberts attempts to brush that law aside in favor of a discussion about constitutional interpretation—“what we’ve been discussing,” he says in line 106, though Kennedy had been less committed to this topic. This resembles Alito’s move in line 50, though Roberts will be less successful than Alito was in forcing Stewart to abandon the recent past in order to partake in constitutional exegesis. “If it’s a book,” Roberts adds, or perhaps begins, for it is not clear whether this is the start of a new hypothetical for which Stewart has to wait, or a reminder that a book was at the center of the hypothetical he sketched in lines 98-100. Stewart does not appear to know either, and waits more than two seconds before repeating “if it’s a book” (ll. 106-107). After the two agree to put aside BCRA’s media exception (which shows Stewart is still thinking about the statute), Stewart takes this as an invitation to fill in the details for himself, sketching (in ll. 113-117) a highly partisan work. Roberts interrupts in line 118 to insist that they return to his hypothetical (from ll. 98-99), involving a single instruction to “vote for X” at the end of a book. Stewart’s turn-initial “yes” in line 124 is not a type-conforming response to a contiguous polar question, but, perhaps, a much-delayed answer to Robert’s surmise in line 100. In any event, the lack of a directly preceding polar question may explain why Stewart feels safe in beginning his answer simply, as it will not be clear what he is saying “yes” to until he explains further. When he does explain, it is to repeat that the corporation would be required to use PAC funds, countering Roberts’s use of “ban” in line 100 but only indirectly and after considerable delay.

In lines 127 and 129, Roberts makes a final attempt to characterize this as a “ban”: “and if they didn’t [use PAC funds] you could ban it.” Now, the “if” clause is crucial, and a significant concession, the first time that one of the conservative justices has acknowledged that Stewart has not claimed for the government the power to ban a book outright. Stewart’s response in lines 130-131 is interesting because rather than simply agree, he restates the entire point (“we could prohibit the publication of the book using the corporate treasury fund”), but only after repeating Roberts’s turn beginning (“if they didn’t”), with the result that Stewart’s response taken as a whole is repetitive: “if they didn’t [use PAC funds rather than general treasury funds] we could prohibit the publication of the book using the corporate [i.e., general] treasury fund.” A possible reason for this redundancy is that Stewart is trying to avoid agreeing with a reflected formulation (of the government’s position) centered on the word “ban.” The problem is that, once again, there is no easy way to disagree with something with which he does not, in fact, disagree, as Roberts did not ask about an outright book ban but one tied to the misuse of general treasury funds. Thus, after an exceptionally long (1.6 seconds) pause, Stewart affects the form of (circuitous) disagreement by offering an alternative to Robert’s formulation that begins the same way (“if they didn’t”) but ends differently, with a restatement of the prohibition against general treasury expenditures.

Not long after the end of Fig. 5, Justice Roberts returned to books, then Justice Souter took up the cause, and then Roberts tried yet again. Without analyzing those exchanges in detail, suffice it to say that they unfolded in a similar fashion, with Stewart explaining that any “prohibition” would merely apply to the use of general treasury funds even as he repeatedly reminded the justices, or attempted to remind them (because he was sometimes interrupted), that printed books were not covered by BCRA even as they were covered by regulations that predated it. In short, Stewart was forced to repeat himself. That he needed to is noteworthy in itself. This conversational circularity, reminiscent of that of decision-makers faced with a vexing problem (Gibson 2012), might be explained by the desire of the conservative justices to drive home a trenchant criticism of Stewart’s position, even at the expense of misrepresenting it, but for the fact that Souter raised the same concern, and he ultimately sided with the government (before the case was reargued, at which point he had retired). Perhaps Alito, Kennedy, and Roberts had succeeded at manufacturing a sense of alarm about a book ban, as evidenced by the fact that someone not on their side took up the argument.

Discussion and Conclusion

“Through artful questioning, Alito, Kennedy, and Roberts had turned a fairly obscure case about campaign-finance reform into a battle over government censorship,” writes Toobin (2013, 166). “The trio made Stewart—and thus the government—take an absurd position, that the government might have the right to criminalize the publication of a 500-page book because of one line at the end.” The foregoing shows how this “artful questioning” was accomplished. Basically, the justices (and especially Alito and Roberts) sketched a hypothetical in which the government, if unchecked by the Court, might someday ban a book, and they worked to keep the specter of this front and center for several minutes. Their conversational resources included the very language of conditional futures (words like “if,” “could,” and “would”) and multiply problematic formulations (a type of declarative question), along with their power to limit Stewart’s responses and exert control over “what we’ve been talking about.” Stewart did what he could under the circumstances, and was given wide latitude to reply with type-nonconforming answers, but his responses were truncated by interruptions and the justices’ perhaps willful obtuseness with regard to the distinction between a funding constraint and an outright “ban,” as a result of which they kept reverting to the latter (or the synonymously employed “prohibit”). They also refused to acknowledge Stewart’s appeals to FECA, on which the “right” to criminalize a book most directly rested. This failure of “uptake” (Austin 1962; Gibson 2000) hinted at their disdain for the complete system of campaign regulation, something that was unmistakable in the Court’s eventual ruling.

Latour (2005, 64) writes that “power, like society, is the final result of a process and not a reservoir, a stock, or a capital that will automatically provide an explanation. Power and domination have to be produced, made up, composed.” For Latour, the implication of this is that the analyst should focus on describing concrete practices, including in the study of law and judging (Latour 2010; also van Oorschot 2021), and resist easy appeals to power that might substitute for that. This is the approach I have taken, mostly eschewing the term except when referring to the power of the government to regulate—the very legal issue in contention. Nonetheless, at this point some observations about the Court’s power are warranted. In oral arguments, the justices’ “situational power” (Collins 2000; also Molotch and Boden 1985) is rooted in a speech-exchange system that allows them to pose hypotheticals, ask the same questions repeatedly, and interrupt at will, all in addition to the fact that they are nine to the advocate’s one and even a small faction of like-minded justices has a numerical advantage when they act in a coordinated fashion. How effectively they wield that power depends partly on their own adroitness and partly on how well the advocate parries, assuming they feel an obligation to attend to his or her responses. This situational power dovetails with several forms of procedural power: to select the cases they hear, to mold the questions considered, to disregard precedent, and to make many important decisions through the “shadow docket,” without even bothering with oral arguments (Vladeck 2023). Of course, all of this only matters inasmuch as the Court’s decisions have consequences. Here things become more uncertain. Hall (2011) presents evidence that Supreme Court decisions reliably modify the behavior of lower courts when, for instance, they dictate what kinds of evidence the latter may admit, but only change the behavior of non-court actors (like Congress and school superintendents) inasmuch as the decisions do not run afoul of public opinion.

Van Oorschot (2018, 233) advocates a perspective in which temporality is not “a backdrop or temporal container of practices” but rather “something that is effectuated within practices, giving rise to multiple and specific ways of making pasts, presents, and futures.” Part of the justices’ power—both during oral arguments and afterward—lies in deciding how much weight to give to the distant past (the Constitution), the recent past (statutes and associated Court cases), the present (the instant case), and the future (as imagined in hypotheticals). In part because of Citizens United, the Roberts Court (or at least the conservative majority) is seen as exceptionally disinterested in stare decisis, or respect for precedent (Segall 2011), preferring to judge statutes in light of an “originalist” reading of the Constitution when it comes to cases about “favored” Constitutional rights like freedom of speech and the right to bear arms. In the initial oral arguments for Citizens, the way that the justices made the Constitution (the distant past) relevant was by imagining, aloud, a particular future in which a book might be banned, though Citizens did not have anything to do with books. Expansive temporality was therefore at play, though it can be debated whether the conservative justices have a genuine fondness for the Constitution or merely invoke it in order warrant decisions that appeal to them for purely ideological reasons (compare Segall 2018 and Wurman 2017).

This research makes three contributions to conversation analysis. First, it extends the study of institutional talk to US Supreme Court oral arguments. Of particular interest from this perspective are the rules of the speech-exchange system, and the resources participants import, perhaps with modification, from ordinary conversation to perform their respective institutional tasks (Drew and Heritage 1992a). In oral arguments, those resources include various ways of talking about possible futures, including through the use of words like “if,” “could,” “would” (Hudson 2008); reflected formulations; and polar questions and techniques for responding evasively to those (Raymond 2003; see also Drew 1990). Again, these conversational resources are asymmetrically available, allowing the justices to dominate interactionally, so long as that domination does not prevent the advocate from eventually having the chance to respond. I also noted some phenomena worthy of further investigation, including consecutive questions without intervening answers, a likely imperative to extract serious import from a seemingly frivolous or facetious question, a flipflopping between “your position” and “the government’s position,” and a kind of lexical appropriation that drove Stewart from “ban” to “prohibit” to “bar.”

Second, by focusing on a setting in which the goal is to persuade, the article contributes to the study of interactional rhetoric. “Rhetoric” is a term conversation analysts almost never use, apart from sometimes characterizing a question as “rhetorical,” though some CA is definitely concerned with relevant practices, such as exaggeration (Drew 2003; Pomerantz 1986). Scholars of rhetoric, in contrast, are very attentive to the work of conversation analysts, though they seem divided between those who whole-heartedly embrace CA (Jackson and Jacobs 1980) and those who maintain a skeptical distance (Cissna and Anderson 2008; Cooren 2008). In the Citizens United exchange, the rhetorical devices included the persistent use of “ban” and its synonym “prohibit,” and the justices’ (especially Kennedy’s) habit of asking questions and offering formulations that were multiply problematic while effectively limiting Stewart to one response each time. Contestation over the topic of conversation—a statute, the Constitution, the Constitution as interpreted by previous Court decisions—is another point of rhetorical interest. Through all of this, it is important to remember the complicated structure of what Goffman (1981) calls the “participation framework”: the advocate addresses the justices in order to persuade them, but sometimes knows that a given justice cannot be persuaded; the justices mostly address the advocate but, it is often said, are “really” trying to test out arguments on one another; and both sides are mindful that their words are available to a much wider audience, and sometimes (it is said) speak mainly with the general public, or the historical record, in mind.

Third, the analysis illustrates some of the advantages of having information about what participants could say, something conversation analysts rarely have access to. How the subfield has been so successful regardless deserves some consideration. One answer is that analysts focus on aspects of speaking turns that display an orientation to sequential placement while taking much of the content of what people say to be a straightforward expression of their past experiences and prior commitments, and therefore beyond CA’s explanatory purview. Another is that analysts assume that people reveal what they have to say when they say it, or at least partly say it (Schegloff 2002), and there is no need to worry about anything else. However, knowledge of what someone is ready, able, or motivated to say is arguably an important analytical resource. First, this draws heightened attention to moments when a response is relevant and available but goes unarticulated, such as when a patient fails to report acute pain during a medical examination (Heritage et al. 2007), or when the president of the USA fails to admit the danger of a proposed course of action when that has been repeatedly explained to him (Gibson 2011a). Second, when a speaker has multiple candidate answers at the ready, and we know what those are, we can ask why one response was selected over another. Third, when participants have advance knowledge of one another’s conversational options, or even just their conversational tendencies (DeLand 2021), they are better able to project what the trajectory of a speaker’s turn might be and respond before it is complete (Jefferson 1984), and perhaps in order to prevent completion (Gibson 2005). For instance, Alito’s interruption of Stewart in line 33 might have been an attempt to prevent him from appealing to FECA, something anticipated in the government’s brief (22), as Alito almost certainly knew. In such situations, it is advantageous for the analyst to have this information as well, for the sake of making sense of the resulting behavior. In sum, information about someone’s conversational capacity, including but not only when that knowledge is shared by other participants, can make an important contribution as the analyst seeks to answer the question “Why that now?” (Bilmes 1985; Schegloff and Sacks 1973) by expanding it to “Why that rather than something else?” and, sometimes, “Why nothing rather than that?”

More generally, the article applies an important sociological lens, both qualitative and disciplined, to a realm that has gone almost untouched by sociologists though its decisions are enormously (if not always straightforwardly) repercussive. Past social-scientific research on oral arguments has been almost entirely quantitative, except for the occasional illustrative quotation. While this is enlightening in its way, it misses justices’ and advocates’ second-by-second engagement with particular legal doctrines and statutes, as well as their tussle over control of the floor and the very topic of conversation. By getting “close to the phenomenon” (Aspers and Corte 2019) we see things that quantitative researchers miss, such as interruptions that prevent an advocate from fully responding to every aspect of a question and type-nonconforming answers that actually go on to agree with the proposition suggested by the justice’s question. And insofar as what gets said in oral arguments is frequently consequential for the Court’s subsequent deliberations and the final decision (Johnson 2004)— the majority opinion in Citizens expressly cites the re-argument transcripts several times—so may be the constitutive practices. Consequently, a full account of any particular case, and its eventual outcome, should include a careful analysis of the oral arguments which serve, from the public’s perspective, as a partly extemporaneous bridge between the written briefs, on the one side, and written rulings, on the other.