Concept
Codefendant Confessions
Right to confront witnesses
Codefendant Confessions
Crawford v. Washington lets the prosecution offer a testimonial out-of-court statement against an accused in only two ways: the declarant testifies at trial and faces cross-examination, or the declarant is unavailable and the accused had an earlier chance to cross-examine.1 A confession to detectives is a core example of a testimonial statement. It narrates a past crime, it is given to government officers, and the prosecution offers it to prove what happened. When the person who confessed is tried beside someone else, the confession raises a confrontation problem of a distinctive kind.
Against the confessor himself, the confession raises no such problem. It is his own statement, admitted as the statement of a party opponent, and he may take the stand to deny or explain it.2 The trouble begins when the same words reach the jury that is deciding a second defendant’s case. If jurors treat the confession as evidence against that defendant, the confessor has functionally become his accuser, and because the confessor is also on trial and will not testify, the accused has no one to cross-examine.3
A joint trial puts the confession before that jury whether or not the judge says it should be set aside. When evidence is admissible against one party and not another, the accepted practice is to admit it with an instruction, if requested, that jurors may consider it only against the party for whom it is admissible.4 Whether that practice works when the evidence is a codefendant’s confession is the question the Bruton rule answers. Severance is the other answer, and motions to sever commonly rest on two grounds: finger-pointing between defendants, and confessions that implicate a codefendant in violation of his right of confrontation.5
The three subsections below keep these facts and vary one thing at a time: first the instruction, then the edit, then whether Ames testifies. On an exam, begin every codefendant problem by asking against whom the confession is offered. Use against the confessor raises no confrontation issue, and use against the other defendant raises all of them.
The Bruton Rule
In Bruton v. United States, Evans’s confession implicating Bruton was admitted at their joint trial. Evans did not take the stand, so Bruton could not cross-examine him.6 The judge told the jury that the confession was competent evidence against Evans but inadmissible hearsay against Bruton, and that it had to be disregarded in deciding Bruton’s guilt.7 The Supreme Court reversed. In the context of a joint trial, it held, a limiting instruction is not an adequate substitute for the defendant’s constitutional right of cross-examination.8 The rule that emerges is this: a nontestifying defendant’s confession implicating a codefendant cannot go to the jury at a joint trial on the strength of an instruction to consider it only against the confessor, although it remains fully usable against the person who made it.
Bruton turned on a premise that the Court declined to keep. The earlier decision in Delli Paoli v. United States had rested on the view that a jury can reasonably follow a sufficiently clear instruction to disregard the confessor’s statement that his codefendant joined him in the crime. If that were true, no confrontation question would arise, because the case would be treated as though the confessor had never implicated the codefendant.9 The Court would not accept that premise for the joint trial.
The holding is deliberately narrow. Courts ordinarily admit evidence of this kind with a limiting instruction, and Bruton marks an exception to that practice.10 The exception covers a setting in which the risk that the jury will not, or cannot, follow the instruction is so great, and the consequence so vital to the defendant, that the practical limits of the jury system cannot be ignored: the powerfully incriminating out-of-court statements of a codefendant who stands accused side by side with the defendant, deliberately spread before the jury in a joint trial.11
Two features of that setting do separate work. The first is force. A confession from an accomplice at the next table is devastating to the person it names, and a jury that has heard it cannot realistically be expected to put it aside. The second is that the accusation cannot be tested. Jurors weighing an accomplice’s testimony are told to be careful because an accomplice has a recognized motive to shift blame onto others, and the unreliability of the account is “intolerably compounded” when the accomplice does not testify and cannot be cross-examined.12 Once the instruction fails, the confessor is functionally the defendant’s accuser, and the defendant has no right to cross-examine him.13
The Court’s later description of Bruton locates the violation in the second feature. The error arose because the declarant “does not testify and cannot be tested by cross-examination,” which implies that no confrontation problem would have existed had Bruton been able to cross-examine his codefendant.14 Bruton put the consequence bluntly: “The effect is the same as if there had been no instruction at all.”15 Justice White, dissenting, would have kept the instruction, reasoning that a codefendant’s confession admitted only against the codefendant cannot compare in evidentiary value with the defendant’s own confession.16
Every cure keeps the confession from the jury that decides the other defendant’s case. Separate trials do it most directly: Ames’s confession is offered only at Ames’s trial, as the statement of a party opponent, and Burke’s jury never hears it.17 The prosecution can instead edit the confession so that it no longer implicates Burke, a technique the next subsection examines. Where the name cannot be removed without pointing at Burke anyway, the remaining choices are severance or giving up the confession.18
Prosecutors sometimes argue that the instruction costs little when the other defendant confessed too and the two confessions interlock. Cruz v. New York rejected that argument and reversed a conviction where a nontestifying codefendant’s incriminating confession was admitted with a limiting instruction alongside the defendant’s own confession.19 A codefendant’s confession that confirms the defendant’s own is more damaging, not less, because a defendant who confessed usually claims that his confession was inaccurately reported or untrue when made, and a second confession telling the same story supports the account he is disputing. Interlocking measures reliability, and reliability neither supplies cross-examination nor shows that jurors will obey an instruction to disregard the statement. After Crawford, reliability cannot stand in for cross-examination of a testimonial statement at all.
Crawford reaches the same result for a confession given to police, which falls within the core class of testimonial statements. As against the other defendant, the confession is testimonial hearsay from a declarant he has never cross-examined, and when the instruction fails, the confessor operates as an adverse witness against him.20 The result does not depend on a joint trial. If the prosecution offers the confession against the nonconfessor at a separate trial, a hearsay exception such as the one for declarations against interest answers only the hearsay question.21 The Clause is a separate hurdle, and confrontation problems arise in just that setting, with an absent declarant and a statement that hearsay law would otherwise admit.22 Unless the declarant is unavailable and the accused had an earlier chance to cross-examine him, the confession stays out.23 The Clause reaches only testimonial statements. Nontestimonial statements are admissible, while testimonial ones without cross-examination are not,24 so an offhand remark by one defendant to a friend implicates the hearsay rules but not the Clause, however plainly it names the other defendant.
The rule binds state courts as well as federal courts, because the Fourteenth Amendment makes the Sixth Amendment’s confrontation guarantee enforceable against the states.25 A violation does not end the case automatically. If the case against the defendant, apart from the codefendant’s confession, was so overwhelming that admitting the confession was harmless beyond a reasonable doubt, the appellate court will not reverse.26 That inquiry comes after the violation question, and it is where the defendant’s own confession can matter, because it is part of the evidence that remains.
Redacted Confessions
The Bruton rule is narrow, and its narrowness gives the prosecution a route. In Richardson v. Marsh, the Court described Bruton as a narrow exception to the assumption that jurors follow their instructions, an exception covering the facially incriminating confession of a nontestifying codefendant introduced at a joint trial.30 A confession that does not incriminate the codefendant on its face falls outside that exception. The prosecution may therefore offer it at the joint trial if it edits out the references to the codefendant, and the edit works unless the jury would obviously infer that the omitted name belongs to the codefendant.31
The line follows the reason for the rule. The confession in Bruton expressly implicated the defendant as the confessor’s accomplice, so there was not the slightest doubt that it would prove “powerfully incriminating” when the jury heard it.32 An edit that removes the accusation removes what made the instruction fail, and courts have generally accepted a redacted statement once it is no longer powerfully incriminating to the defendant.33 The ordinary assumption that jurors follow their instructions then governs again.34
In Richardson, Marsh and Williams were tried jointly, and the State introduced Williams’s confession, edited to omit all reference to Marsh and, indeed, all indication that anyone other than Williams and a man named Martin took part in the crime. When the confession was admitted, the jury was admonished not to use it against Marsh.35 Williams did not testify. The Court held that the confession fell outside Bruton, because it was not incriminating on its face and became so only when linked with evidence introduced later at trial, namely Marsh’s own testimony.36
The tempting argument that Bruton applies because other evidence at trial ties Marsh to the events the confession describes therefore fails. The test looks at the confession as the jury hears it, read by itself, and Richardson placed a confession that incriminates only by inference from later evidence outside the exception.37
An edit that is too light fails in the opposite direction. In Gray v. Maryland, the redaction substituted the word “deleted” for the codefendant’s name, and the Court reversed the conviction because the jury could too easily see what the word stood for.38 A confession that reads, “Me, deleted, deleted, and a few other guys killed the victim,” announces with each blank that a name was taken out, and jurors who can see who else sits at the defense table will put the name back. Deleting a name is constitutionally defective when the jury is certain to infer that the omitted name is the codefendant’s.39 The accusation reaches the jury in almost the form it had before the edit, so the instruction fails for the same reason it failed in Bruton.
Samia v. United States supplies an edit that works. There the Court upheld the admission, at a joint trial, of a nontestifying codefendant’s confession edited to describe the defendant only as “the other person,” with an instruction to consider the confession against the confessor alone. The decision fits the practice lower courts had already accepted, substituting a neutral term for the defendant’s name, because the edited statement is no longer powerfully incriminating.40 Samia permits a confession that admits another participant existed, so long as the statement does not supply that participant’s identity.
Gray and Samia both remove the name, so the missing name cannot be what separates them. What separates them is what remains on the page. A blank or the word “deleted” shows that a particular name was taken out, while a neutral description reads as an ordinary account of another participant whose identity the statement does not supply. Richardson sits further along the same line, since the confession there gave no sign that anyone besides the confessor and Martin was involved. In Richardson and Samia, the jury was also told to consider the confession only against the confessor, so the instruction works alongside the edit. When the reference to the codefendant cannot be removed without leaving the confession pointing at him, the remaining choices are severance or giving up the confession.41
Testifying Confessors
The Bruton problem exists only because the confessor stays silent. Crawford states the converse in general terms: when the declarant appears for cross-examination at trial, the Confrontation Clause places “no constraints at all” on the use of his prior testimonial statements.43 A codefendant’s custodial confession is testimonial, and its author is the accuser whom the other defendant must be able to test. If the author takes the stand, that accuser is in the courtroom, under oath, answering questions. The Court’s later description of Bruton drew the same inference: the error there arose because the declarant “does not testify and cannot be tested by cross-examination,” which implies that no confrontation problem would have existed had Bruton been able to cross-examine his codefendant.44
The reason follows from the two features that made the Bruton rule necessary. A confession from an accomplice sitting at the next table is devastating to the person it names, and that person ordinarily cannot question the accuser. When the confessor testifies, the second half of the problem disappears. Jurors weigh an accomplice’s testimony carefully because an accomplice has a recognized motive to shift blame onto others,45 and counsel for the named defendant can now probe that motive, along with what the accomplice perceived and how well he remembers. Because the confessor can be cross-examined about both the earlier statement and the present testimony, the Clause is satisfied.46
What the confessor says once on the stand does not decide the matter, and Nelson v. O’Neil shows why. A court might expect effective confrontation of a confessing accomplice to require that he stand by the statement. In Nelson, the codefendant took the stand, denied making the confession, asserted that its substance was false, and testified favorably to the defendant. The Court held that the defendant “has been denied no rights protected by the Sixth and Fourteenth Amendments,” and it reasoned that the denial put him in a “more favorable” situation than an affirmation would have.47 A confession its author repudiates under oath asks the jury to credit a version the author has abandoned, and the other defendant’s lawyer may question the author about both versions. The Clause asks whether the accuser can be questioned, not which story the accuser tells, so an admission, a denial, and a partial retraction each leave the defendant with the confrontation the Clause guarantees.48
Nelson states the usual result, not a guarantee, because the rule depends on cross-examination that is actually available. Two defendants who share a lawyer present one danger. The lawyer can hardly cross-examine one client effectively without discrediting the other, so a decision not to cross-examine, or to do so only pro forma, would violate the Bruton rule. The prosecution creates another. If the confessor disavows the confession and the prosecutor then introduces the unredacted confession to impeach the disavowal, that too would violate the rule, which prohibits any indirect use of an unredacted confession against a codefendant.49 The impeachment is aimed at the confessor, yet the jury hears an accusation naming the other defendant and is again asked to confine it to the confessor, which is the instruction Bruton refused to trust. Cross-examination also has a cost that separate trials avoid: it can highlight evidence that is inadmissible against the other defendant, and antagonism between defendants who question each other may itself make severance necessary.50
The opportunity to cross-examine the confessor can also come from an earlier proceeding. It has been held sufficient even where the confessor did not testify at the criminal trial,51 an approach with roots in Green, which admitted preliminary-hearing testimony given under circumstances closely approximating those of a typical trial. Language in Green suggests that any earlier opportunity to cross-examine would do, but the Court has since cautioned that it has not decided what follows when there was no questioning, or only de minimis questioning, at the earlier proceeding. The earlier opportunity must also reach the right subject. Lee v. Illinois rejected the argument that a suppression hearing was enough, because that hearing decides whether the confession was voluntary, a question on which the truth of the statement is irrelevant, so there was no opportunity to cross-examine the maker about its reliability.52
A codefendant’s confession can also reach the jury without a testifying confessor when it is offered for a purpose other than its truth. Cross-examination tests assertions, and a statement offered for another purpose asks the jury to credit nothing the declarant said. Crawford confirms that the Clause does not bar a testimonial statement used for a purpose other than establishing the truth of the matter asserted,53 and it relied on Tennessee v. Street, where a codefendant’s confession was offered, with an instruction limiting its use, to rebut the defendant’s claim that his own confession was coerced. Two cautions keep this route narrow. The not-for-truth label has to be genuine, as the earlier treatment of expert witnesses showed. And a trial judge keeps the power to exclude such a statement under Rule 403 when, on its face, it asserts facts directly relevant to a critical issue that the declarant presumably knew, because jurors then run a grave risk of using it as substantive proof.54
On an exam, sort a codefendant-confession stem by asking whether the confessor can be questioned about the facts the confession asserts. A confessor who testifies, whether to admit, deny, or qualify the statement, removes the confrontation objection, so a stem that stresses the confessor’s denial is testing the Nelson point, and the denial changes nothing. A confessor who stays off the stand, perhaps by invoking the privilege against self-incrimination, leaves the Bruton analysis in place. A complete answer also checks whether the confession is offered for a genuine nonhearsay purpose, and whether shared counsel or an unredacted impeachment use has emptied the cross-examination of a confessor who does testify.