Opinion
NO. PD-0196-16
02-08-2017
RODNEY DIMITRIUS LAKE, Appellant v. THE STATE OF TEXAS
ON STATE'S PETITION FOR DISCRETIONARY REVIEW FROM THE SECOND COURT OF APPEALS TARRANT COUNTY
KELLER, P.J., announced the judgment of the Court and delivered an opinion in which KEASLER, HERVEY and RICHARDSON, JJ., joined. YEARY, J., filed a concurring opinion in which NEWELL and KEEL, JJ., joined. ALCALA, J., filed a dissenting opinion. WALKER, J., concurred.
We must decide whether the denial of closing argument at a community-supervision revocation proceeding is the sort of error that is exempt from a harm analysis. We conclude that it is not, because the Supreme Court has not labeled it as structural. Consequently, we reverse the judgment of the court of appeals and remand the case for a harm analysis.
I. BACKGROUND
A. Trial
A jury found appellant guilty of sexual assault of a child under age seventeen and assessed a sentence of ten years in prison and a $10,000 fine. Upon the recommendation of the jury, sentence was suspended and appellant was placed on community supervision for ten years.
Prior to September 1, 1993, community supervision was referred to as probation. See Yazdchi v. State, 428 S.W.3d 831, 833 n.1 (Tex. Crim. App. 2014). We use the terms interchangeably in this opinion. See id.
The State later filed a motion to revoke community supervision. Appellant pled "not true" to the allegations in the motion. After the parties presented testimony, the following occurred:
[DEFENSE COUNSEL]: Defense rests.
THE COURT: Rest?
[DEFENSE COUNSEL]: Yes, Your Honor.
[PROSECUTOR]: State will close.
[DEFENSE COUNSEL]: Defense closes, Your Honor. Can we make a closing statement when the time comes, Your Honor?
THE COURT: Sir?
[DEFENSE COUNSEL]: Can I make a closing statement when the time comes?
[THE COURT]: I don't need one.
The trial court then found allegations one, two, and five to be "not true" and allegations three and four to be "true." The court revoked appellant's community supervision and imposed the previously assessed sentence.
B. Appeal
One of appellant's complaints on appeal was that the trial court erred by refusing to allow defense counsel to make a closing argument. The State responded that appellant had failed to preserve error. The court of appeals addressed and rejected the State's contention with respect to preservation. Then, citing Herring v. New York and other cases, the court of appeals concluded that the trial court had violated appellant's Sixth Amendment right to the effective assistance of counsel and his state constitutional right to be heard. In a section entitled "Reversible Error Presumed from Denial of Closing Argument," the court of appeals concluded, without elaboration, that, "[b]ecause the error is constitutional and the effect of the denial of closing argument cannot be assessed, the error is reversible without any showing of harm." The court of appeals reversed the trial court's revocation judgment and remanded the case for a new trial on revocation.
Lake v. State, 481 S.W.3d 656, 658-60 (Tex. App.-Fort Worth 2015). The State has not sought review of the preservation holding. We express no opinion on whether error was actually preserved.
422 U.S. 853 (1975).
Lake, 481 S.W.3d at 660.
Id. at 661.
II. ANALYSIS
A. State's Complaint
The State complains that the court of appeals erred in treating the refusal to allow closing argument as "structural error immune from a harmless error analysis" and that the court of appeals's decision "is contrary to decisions of the United State's Supreme Court and this Court defining what constitutes structural error." The State argues that the United States Supreme Court "has not held that a probationer has a constitutional right to closing argument, much less that any such right to argument is structural." The State claims that, while this Court "seems to have recognized" a state constitutional right to present closing argument in community-supervision revocation proceedings, under Black v. Romano and Gagnon v. Scarpelli, there is no federal constitutional right to present closing argument in community-supervision proceedings, and absent a federal constitutional right, there can be no structural error. Relying upon a footnote in Herring v. New York, the State also suggests that even the denial of a federal constitutional right to present closing argument is not structural error or, at least, not with respect to community-supervision revocation proceedings. The State further contends that the record in this case contains sufficient data upon which to assess whether the denial of closing argument was harmful, and the State concludes, after discussing the record, that the error was harmless beyond a reasonable doubt.
The State's grounds for review were as follows:
1. The court of appeals erred in treating the trial court's refusal to allow final argument before revoking Appellant's community supervision as structural error immune from a harmless error analysis.
2. The court of appeals' treatment of the trial court's refusal to allow final argument before revoking Apellant's community supervision as structural error immune from a harmless-error analysis is contrary to decisions of the United States Supreme Court and this Court defining what constitutes structural error.
471 U.S. 606 (1985).
411 U.S. 779 (1973).
422 U.S. at 863 n.13.
B. Standard for Determining Structural Error
In Cain v. State, we issued a broad mandate that nearly all errors would be subject to a harm analysis, with only limited exceptions as follows: "Except for certain federal constitutional errors labeled by the United States Supreme Court as 'structural,' no error . . . is categorically immune to a harmless error analysis." As the standard suggests, only federal constitutional errors can be "structural," though most federal constitutional errors are not structural. Subsequent cases have reaffirmed that we treat error as "structural" only if the Supreme Court has labeled it as such.
Gonzales v. State, 994 S.W.2d 170, 171(Tex. Crim. App. 1999) (quoting Cain v. State, 947 S.W.2d 262, 264 (Tex. Crim. App. 1997)).
Schmutz v. State, 440 S.W.3d 29, 35 (Tex. Crim. App. 2014).
Mercier v. State, 322 S.W.3d 258, 263 (Tex. Crim. App. 2010) ("The court of appeals overlooked Cain's holding that only errors labeled as structural by the Supreme Court are immune from a harm analysis."); Williams v. State, 273 S.W.3d 200, 225 (Tex. Crim. App. 2008) ("The erroneous submission of an unwanted defensive issue has not been labeled by the United States Supreme Court as structural.").
Even when an error that is not structural under Cain seems to defy proper analysis or the data seems to be insufficient to assess harm, an appellate court is "obligated to conduct a thorough analysis to determine the extent of harm caused by this error before reversing the conviction." For federal constitutional error that is not structural, the applicable harm analysis requires the appellate court to reverse unless it determines beyond a reasonable doubt that the error did not contribute to the defendant's conviction or punishment. If, after such analysis, the harm of the error simply cannot be assessed, then "the error will not be proven harmless beyond a reasonable doubt," but "appellate courts should not automatically foreclose application of the harmless error test."
Gonzales, 994 S.W.2d at 171.
TEX. R. APP. P. 44.2(a). See Chapman v. California, 386 U.S. 18, 24 (1967).
Cain, 947 S.W.2d at 264. See also Gonzales, 994 S.W.2d at 171-72.
C. Not Labeled Structural
1. The Error at Issue is Herring Error
In Herring v. New York, the Supreme Court held that the right to the assistance of counsel guaranteed by the Sixth Amendment was violated when a trial court refused to allow counsel to make a closing argument at the guilt phase of trial. Prior precedent from this Court indicates that the refusal to allow defense counsel to make a closing argument at a community-supervision revocation proceeding is Herring error.
Ruedas v. State, 586 S.W.2d 520, 524 (Tex. Crim. App. 1979) (citing Herring and Ex parte Flores, 537 S.W.2d 458 (Tex. Crim. App. 1976)). See Ex parte Shivers, 501 S.W.2d 898, 900-01 (Tex. Crim. App. 1973) (holding that Mempa v. Rhay, 389 U.S. 128 (1967) controlled with respect to Texas revocation proceedings that suspended the imposition of sentence and accorded a constitutional right to counsel and further holding that Gagnon v. Scarpelli was inapplicable); Flores, 547 S.W.2d at 459 (holding it to be "well settled that criminal defendants are entitled to counsel at probation revocation proceedings under the Sixth Amendment" and citing Mempa and Shivers). See also Ex parte Doan, 369 S.W.3d 205, 209-10 (Tex. Crim. App. 2012) (Texas probation revocation system is "judicial and adversarial," providing "trial-like proceedings," in which "the State is represented by a prosecutor, the defendant does have a right to counsel, the hearing is before a judge, formal rules of evidence do apply, and there may be appeal directly to a court of appeals" and consequently, "[a] community supervision revocation proceeding in Texas bears little resemblance to the administrative hearing described in Scarpelli.").
2. Herring Did Not Label the Error Structural
Herring addressed only whether a constitutional violation occurred; it did not address whether to apply a harm analysis, and it did not cite Chapman v. California, the seminal case establishing the federal constitutional harm standard. Concluding that New York denied the defendant "the assistance of counsel that the Constitution guarantees," the Court vacated the case and remanded "for further proceedings not inconsistent with this opinion." Herring made a number of statements about the importance of closing argument, and the Court suggested that it was not appropriate for the trial court to determine whether argument could potentially affect the outcome of the proceedings. But none of these statements, which were all about whether a trial judge erred in failing to permit closing argument, speak to whether the error is immune from a harm analysis. So, even recognizing that Herring was decided before the term "structural" came into usage, nothing in that opinion can be construed as labeling the error it describes as structural.
386 U.S. 18 (1966).
See Herring, 422 U.S. at 856-865.
Id. at 865.
Id. at 858 ("There can be no doubt that closing argument for the defense is a basic element of the adversary factfinding process in a criminal trial."), 862 (Closing argument "serves to sharpen and clarify the issues for resolution" and "[i]n a criminal trial, which is in the end basically a factfinding process, no aspect of such advocacy could be more important than the opportunity finally to marshal the evidence for each side before submission of the case to judgment."), 863 ("[T]he difference in any case between total denial of final argument and a concise but persuasive summation could spell the difference, for the defendant, between liberty and unjust imprisonment.").
Id. at 858 ("[C]ounsel for the defense has a right to make a closing summation to the jury, no matter how strong the case for the prosecution may appear to the presiding judge."), 864 ("Some cases may appear to the trial judge to be simple—open and shut—at the close of the evidence" and "surely in many such cases a closing argument will, in the words of Mr. Justice Jackson, be 'likely to leave [a] judge just where it found him.'" But, "there will be cases where closing argument may correct a premature misjudgment and avoid an otherwise erroneous verdict" and "there is no certain way for a trial judge to identify accurately which cases these will be, until the judge has heard the closing summation of counsel.").
See Arizona v. Fulminante, 499 U.S. 279, 309-10 (1991) (listing situations that involve a "structural defect" as being immune from a harm analysis).
This is true even though several of the federal circuits have read Herring as saying that error in denying closing argument is immune from a harm analysis. The Supreme Court has only once come close to expressly addressing whether any of these lower courts are correctly reading Herring on this issue. In Glebe v. Frost, addressing the Ninth Circuit's contention that the denial of closing argument is structural error, the Supreme Court skirted the question by distinguishing a restriction on argument from the complete denial of argument: "According to the Ninth Circuit, Herring further held that this denial amounts to structural error. We need not opine on the accuracy of that interpretation. For even assuming that Herring established that complete denial of summation amounts to structural error, it did not clearly establish that the restriction of summation also amounts to structural error." Given the Supreme Court's silence on the matter when the issue was argued, and our own reading of Herring, we cannot agree with those courts that hold that Herring labeled the error as structural.
United States v. Davis, 993 F.2d 62, 64 (5th Cir. 1993) ("Given the difficulty of determining the prejudicial impact of the failure to afford summation, the denial of a request for it is reversible error per se.") (citing Herring, 422 U.S. at 864); Patty v. Bordenkircher, 603 F.2d 587, 589 (6th Cir. 1979) (saying that trial court erred in applying a harm analysis because the Herring case "indicated that the strength of the prosecution's case is not a factor"); United States v. Spears, 671 F.2d 991, 992 (7th Cir. 1982) ("In Herring v. New York . . . , the Supreme Court held that it is per se reversible error in any criminal trial, whether to a jury or to a judge, for the trial court to deny the defendant the opportunity to present a closing argument."); Frost v. Van Boening, 757 F.3d 910, (9th Cir.), rev'd sub nom., Glebe v. Frost, 135 S. Ct. 429 (2014) ("Herring error is structural.") (citing Herring, 422 U.S. at 864-65). See also United States v. King, 650 F.2d 534, 536 (4th Cir. 1981) (citing Chapman v. California and stating that the denial of the right to closing argument "is not harmless error."); Thomas v. United States, 473 A.2d 378 (D.C. App. 1984) ("The denial of that right [in Herring] can never be deemed harmless."); State v. Gilman, 489 A.2d 1100, (Me. 1985) ("In the absence of waiver, the denial of the right to present closing argument 'must result in reversal of the conviction without regard to whether the defendant was prejudiced.'") (emphasis in original). But see United States v. Arbolaez, 450 F.3d 1283, 1294-95 (11th Cir. 2006) (suggesting that Herring error that occurs at the forfeiture stage of a federal criminal trial might be structural or might be subject to the federal constitutional harmless error standard).
135 S. Ct. 429, 431 (2014) (emphasis in original).
2. Subsequent Supreme Court Decisions Did Not Label the Error Structural
If Herring does not assign the "structural" label to the error, do subsequent Supreme Court cases? The Supreme Court has stated that only "a very limited class" of errors is structural, and it has not included Herring error when it has listed examples of structural error.
United States v. Davila, 133 S. Ct. 2139, 2149 (2013); United States v. Marcus, 560 U.S. 258, 263 (2010); Neder v. United States, 527 U.S. 1, 8 (1999); Johnson v. United States, 520 U.S. 461, 468 (1997).
The total denial of counsel at trial is listed as an example of structural error, but the Court has said that a violation of the right to counsel is "subject to harmless-error analysis . . . unless the deprivation, by its very nature, cannot be harmless." A violation of the right to counsel that merely affects the admission of evidence, for example, is subject to a harm analysis. To be immune from a harm analysis, a violation of the right to counsel must "pervade the entire proceeding." Whether the denial of closing argument at a stage of trial is the sort of error that "pervades the entire proceeding" is a question the Supreme Court has not answered.
See supra nn. 23, 26.
Rushen v. Spain, 464 U.S. 114, 117 n.2 (1983).
Satterwhite v. Texas, 486 U.S. 249, 257 (1987).
Id. See also Penson v. Ohio, 488 U.S. 75, 88 (1988) ("[A] pervasive denial of counsel casts such doubt on the fairness of the trial process, that it can never be considered harmless error.").
Appellant claims that Herring error was recognized as exempt from a harm analysis in United States v. Cronic. In Cronic, the Supreme Court cited Herring as an example of when prejudice is presumed for Sixth Amendment purposes. Prejudice is a component of an ineffective assistance of counsel violation under the Sixth Amendment. To even establish the existence of constitutional error as a result of ineffective assistance, a defendant must generally show prejudice. But there are some situations involving the absence of counsel or government interference with counsel's representation in which a showing of prejudice is not required, where the constitutional violation is established without it, and in those cases (such as Herring) prejudice is "presumed."
466 U.S. 648, 659 n.25 (1984). He also cites Woods v. Donald, which recites several of Cronic's statements that we discuss below. See 135 S. Ct. 1372, 1375-76, 1378 (2015).
Cronic, 466 U.S. at 659 n.25 ("The Court has uniformly found constitutional error without any showing of prejudice when counsel was either totally absent, or prevented from assisting the accused during a critical stage of the proceeding.") (citing Herring among other cases). See also Bell v. Cone, 535 U.S. 685, 696 n.3 (2002) ("In a footnote, we also cited other cases . . . where we found a Sixth Amendment error without requiring a showing of prejudice. Each involved criminal defendants who had actually or constructively been denied counsel by government action.") (discussing Cronic and citing Herring among other cases); Strickland v. Washington, 466 U.S. 668, 692 (1984) ("In certain Sixth Amendment contexts, prejudice is presumed. Actual or constructive denial of the assistance of counsel altogether is legally presumed to result in prejudice. So are various kinds of state interference with counsel's assistance.") (citing Cronic, 466 U.S. at 659 n.25).
Strickland, 466 U.S. at 692.
Id. at 691-92.
Cone, 535 U.S. at 696 n.3; Cronic, 466 U.S. at 659 n.25; Strickland, 466 U.S. at 692.
Is that presumption rebuttable? To conclude on the basis of Cronic and other "presumption of prejudice" cases that Herring error has been labeled "structural," we would have to hold that the Supreme Court views the presumption as never rebuttable. That is, we would have to hold that the Supreme Court views a declaration that prejudice is presumed with respect to a certain type of error as the same thing as saying that that type of error is structural. Some federal circuits have made statements that seem to equate these concepts. However, some of these circuits, and others, have suggested that these concepts are not entirely coextensive—that is, that some of these errors in which prejudice is presumed may nevertheless be subject to a harm analysis.
United States v. Ragin, 820 F.3d 609, 617-18 (4th Cir. 2016) ("The Court stated that in certain limited contexts, 'prejudice is presumed' . . . . That a case warrants a finding of presumed prejudice under any of these three prongs is an extremely high showing for a criminal defendant to make. . . . If, however, the defendant makes such a showing, it would necessarily follow that there was a structural error.") (some internal quotation marks omitted); United States v. Roy, 761 F.3d 1285, 1293 (11th Cir. 2014) ("We, however, have previously held that 'structural error exists where counsel is "prevented from assisting the accused during a critical stage of the proceeding."'") (citing Arbolaez); Arbolaez, 450 F.3d at 1294 ("Finally, constructive denial of counsel is 'legally presumed to result in prejudice' and thus to constitute a structural error."); French v. Jones, 332 F.3d 430, 438 (6th Cir. 2003) ("Despite the appellant's attempt to characterize Cronic's language as dicta, the Court has often held, both before and after Cronic, that absence of counsel during a critical stage of a trial is per se reversible error.").
Sweeney v. United States, 766 F.3d 857, 861 & n.3 (8th Cir. 2014) ("[S]everal of our sister circuits have stated post-Cronic that, under some circumstances, harmless error analysis may apply for an absence of counsel at a critical stage.") (citing cases); Arbolaez, 450 F.3d at 1295 ("We cannot call this error harmless beyond a reasonable doubt as to this phase of the trial."); United States v. Morrison, 946 F.2d 484, 503 n.6 (7th Cir. 1991) ("We emphasize the distinction between an ineffective assistance analysis and a harmless error analysis. In the first, we determine whether a Sixth Amendment violation has occurred. Depending on whether Strickland or Cronic applies, prejudice may or may not be a critical component of such an analysis. In the second, we determine whether the constitutional violation is harmless. The fact that we might presume prejudice for the purposes of the former inquiry does not necessarily dictate that we do the same for the purposes of the latter."); Green v. Arn, 809 F.2d 1257, 1263 (6th Cir. 1987) ("Where the Sixth Amendment claim is the denial, rather than the ineffective assistance, of counsel, the criminal defendant need only show that counsel was absent during a critical stage of the proceedings in order to establish the constitutional violation. Absence from the proceedings is deficient performance as a matter of law, and prejudice is presumed. Nonetheless, a harmless error analysis is appropriate in some instances."); Siverson v. O'Leary, 764 F.2d 1208, (7th Cir. 1985) ("[A] defendant need not affirmatively prove prejudice under the second prong of the Strickland test in order to establish a Sixth Amendment violation based on the lack of defense counsel's assistance at a critical stage of the criminal proceedings. We hold instead that the proper standard for determining prejudice resulting from the erroneous absence of Siverson's counsel during jury deliberations and the return of the verdict is the same standard that was applied to similar errors prior to Strickland: whether the error was harmless beyond a reasonable doubt under Chapman v. California.").
The fact that the Supreme Court uses both terms would seem to indicate that they are not intended to mean the same thing. The Supreme Court has suggested some linkage between presuming prejudice and an error being structural, but it has not said that these categories are coextensive, despite the fact that it has several times given lists of examples of structural error. With respect to cases in which prejudice is presumed, the Supreme Court has said, "Prejudice in these circumstances is so likely that case-by-case inquiry into prejudice is not worth the cost." The Court has also said that "a trial is unfair if the accused is denied counsel at a critical stage of his trial." But these statements were made in the context of assessing whether a constitutional violation had been shown and evince merely that, in certain circumstances, the Supreme Court does not wish to impose the burden of showing harm upon defendants-a burden that would otherwise be required by Strickland. That does not preclude the presumption of harm being rebutted, if the record shows the harmlessness of the error to be obvious, which would have to be the case to satisfy the "beyond a reasonable doubt" standard articulated in Chapman. Because the Supreme Court has not said that errors are always structural when prejudice is presumed, we disagree with appellant's argument that calling Herring error a type of error in which prejudice is presumed is the same as labeling Herring error as structural.
Bank of Nova Scotia, 487 U.S. 250, 256-57 (1988) ("We think, however, that an alternative and more clear explanation is that these cases are ones in which structural protections of the grand jury have been so compromised as to render the proceedings fundamentally unfair, allowing the presumption of prejudice.") (citing Vasquez v. Hillery, 474 U.S. 254 (1986)). See Neder, 527 U.S. at 8 (citing Vasquez as a case involving structural error).
See supra nn.23, 26.
Strickland, 466 U.S. at 692. See also Cronic, 466 U.S. at 658 ("There are, however, circumstances that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.").
Cronic, 466 U.S. at 659 (citing Herring among other cases in footnote).
See Mickens v. Taylor, 535 U.S. 162, 166 (2002) ("As a general matter, a defendant alleging a Sixth Amendment violation must demonstrate 'a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different.' . . . We have spared the defendant the need of showing probable effect upon the outcome, and have simply presumed such effect, where assistance of counsel has been denied entirely or during a critical stage of the proceeding. When that has occurred, the likelihood that the verdict is unreliable is so high that a case-by-case inquiry is unnecessary.") (citing Strickland after the first sentence and Cronic after the last sentence).
See Perry v. Leeke, 488 U.S. 272, 277 n.2 (1989) (recognizing some lower court cases that found a Sixth Amendment violation under Geders v. United States, 425 U.S. 80 (1976)—one of the cases cited along with Herring in Cronic's footnote—to be harmless). See also Cronic, 466 U.S. at 659 n.25 (citing Geders, Herring, and other cases).
We are aware of Holloway v. Arkansas, which said, "[T]his Court has concluded that the assistance of counsel is among those 'constitutional right so basic to a fair trial that their infraction can never be treated as harmless error.' Accordingly, when a defendant is deprived of the presence and assistance of his attorney, either throughout the prosecution or during a critical stage in, at least, the prosecution of a capital offense, reversal is automatic." 435 U.S. 475, 489 (1978) (citation to Chapman omitted; also citing Gideon v. Wainwright, 372 U.S. 335 (1963); Hamilton v. Alabama, 368 U.S. 52 (1961); White v. Maryland, 373 U.S. 59 (1963)). We point out that appellant in this case was not deprived of the presence of his attorney, nor is this case a capital case. In any event, the Supreme Court has later retreated from Holloway's broad language, explaining that Holloway "creates an automatic reversal rule only where defense counsel is forced to represent codefendants over his timely objection," Mickens, 535 U.S. at 168, and that a defendant's reliance on Holloway was "misplaced" because "Gideon, Hamilton, and White were all cases in which the deprivation of the right to counsel affected—and contaminated—the entire criminal proceeding," Satterwhite, 486 U.S. at 257.
We conclude that the Supreme Court has not labeled Herring error as structural—even at the guilt stage of trial, much less on revocation. This holding is consistent with our policy to generally require a harm analysis—precluding one only where the Supreme Court has expressly eschewed it—and it is the cautious approach, because the constitutional harm standard is an onerous one for the State, and the nature of the error can be considered in determining whether harmlessness is in fact shown beyond a reasonable doubt. We reverse the judgment of the court of appeals and remand the case to that court to conduct a harm analysis. Delivered: February 8, 2017
Publish
At the punishment stage of trial, there is at least one situation in which the denial of closing argument would seem to be obviously harmless: when a defendant who is not eligible for probation receives the minimum possible punishment. This scenario may seem less persuasive, however, when one considers that even the total denial of counsel at the punishment stage would seem harmless in that situation. Perhaps even error that is labeled "structural" can actually be harmless if the defendant "wins" at the relevant stage of trial. But since he would be unlikely to complain of receiving the minimum punishment, it is unlikely that we will ever address that apparent anomaly.
Revocation hearings involve at least one factor not present in the original trial that may make such cases more amenable to a harm analysis: a defendant may have been subject to revocation more than once, so that the trial judge may have already heard the possible arguments that could be made in the defendant's favor.
See Snowden v. State, 353 S.W.3d 815, 822 n.31 (Tex. Crim. App. 2011) ("There is no set formula for conducting a harm analysis that necessarily applies across the board, to every case and every type of constitutional error. Perhaps the most that can be said is that what Harris called the 'nature' of the constitutional error will frequently suggest, and sometimes dictate, other relevant considerations in an appropriate harm analysis.").
The concurring opinion believes that the categorical language in Cain "may not have survived the enactment of Rule 44.2(a)" because that rule contains the phrase "[i]f the appellate record in a criminal case reveals constitutional error that is subject to harmless error review." We disagree. The new phrase in Rule 44.2(a) is simply a recognition, consistent with Cain, that the Supreme Court has dictated that certain federal constitutional errors that it has labeled as "structural" are exempt from a harm analysis. At the time Cain was decided (June 18, 1997), this Court had already submitted the current version of Rule 44.2(a) for public comment. See Approval of Revisions to the Texas Rules of Appellate Procedure, 60 Tex. Bar J. 408, 454 (May 1997). The rule was finally approved along with other changes in the rules of appellate procedure on August 15, 1997, to take effect on September 1. Court of Criminal Appeals, Final Approval of Revisions to the Texas Rules of Appellate Procedure, TEX. R. APP. P. (West 1998). In Cain, we reasoned, in part, "The plurality [in Matchett v. State, 941 S.W.2d 922 (Tex. Crim. App. 1996)] . . . reasoned that appellate courts should not foreclose entire categories of error from harmless error review merely because such errors may generally resist a meaningful harmless error determination. . . . We find the plurality opinion in Matchett to be well-reasoned, and we adopt that opinion's holding and reasoning. Except for certain federal constitutional errors labeled by the United States Supreme Court as 'structural,' no error, whether it relates to jurisdiction, voluntariness of a plea, or any other mandatory requirement, is categorically immune to a harmless error analysis." 947 S.W.2d at 264. This statement from Cain reads as a broad statement of new law, not as an interim construction of a rule that will be replaced in a few months time. In Carranza v. State, we stated that "[a] careful reading of Rule 44.2 and our relevant case law reveals several types of error that can occur in criminal cases" with one of those types being "constitutional error that is not subject to a harmless error analysis (i.e. structural error)," and in support of that proposition we cited Cain. 980 S.W.2d 653, 656 & n. 3 (Tex. Crim. App. 1998). In Cena v. State, we confronted a court of appeals's refusal to conduct a harm analysis regarding a trial court's error in refusing to allow a proper question in voir dire. 991 S.W.2d 283 (Tex. Crim. App. 1999) (May 12). On the basis of Cain, we summarily reversed the court of appeals's decision and remanded the case for a harm analysis. Id. at 283. Although the court of appeals's decision was handed down two days before Rule 44.2(a) became effective, see Cena v. State, 960 S.W.2d 804 (Tex. App.-El Paso 1997) (August 29), we would, before our decision in Cena, ratify the application of Rule 44.2 to appeals that were pending when the rule was adopted. Fowler v. State, 991 S.W.2d 258, 260 (Tex. Crim. App. 1999) (March 31). In Gonzales v. State, we addressed another case in which the defense had been denied the right to ask a proper question in voir dire—which we acknowledged was a violation of the Texas Constitution. 994 S.W.2d 170, 171-72 (Tex. Crim. App. 1999). Although our prior decision in Nunfio v. State, 808 S.W.2d 482 (Tex. Crim. App. 1991) had held that such an error was not subject to a harm analysis, we held that Nunfio was no longer controlling after Cain. Gonzales, 994 S.W.2d at 171. We further explained that Cain "issued a broad mandate that overruled several prior cases" and repeated our statement that "appellate courts should not automatically foreclose the application of the harmless error test to certain categories of errors." Id. at 172. In our unanimous decision in Gray v. State, we were emphatic about the categorical nature of Cain's holding and its continued application:
In many other cases we have reaffirmed that we meant what we said—sometimes summarily reversing intermediate appellate decisions when the court of appeals refused to conduct a harm analysis. We have recognized that Cain issued a "broad mandate" that brought the overwhelming majority of errors within the purview of a harm analysis. We have implemented the holding in Cain in a variety of statutory contexts, including the denial of severance, proceeding to trial with eleven jurors, the denial of individual voir dire, and the failure to comply with a statutory requirement to admonish the defendant on the record regarding the full range of punishment. We have made it clear that an error that is not of federal constitutional dimension cannot be deemed "structural." And even when an error does constitute a violation of the federal constitution, the error is "structural" only if the Supreme Court has labeled it as such.159 S.W.3d 95, 96-97 (Tex. Crim. App. 2005). --------