OPINION OF THE COURT
ALITO, Circuit Judge:The State of New Jersey has appealed from two orders of the district court granting Gabriel Pemberthy’s and Rigoberto Monca-da’s petitions for writs of habeas corpus. The district court held that the prosecutor at the petitioners’ joint trial violated their equal protection rights by peremptorily challenging five prospective jurors. After conducting a hearing regarding the prosecutor’s reasons for these challenges, the district court held that the prosecutor had dismissed the jurors in question simply because they were Latinos and not because of a sincere concern about their ability as Spanish speakers to accept the translation of tape-recorded conversations offered in evidence. In the alternative, the court broadly held that dismissing Latino jurors because they can speak Spanish is tantamount to dismissing them based on race and is thus unconstitutional.
We hold that the district court erred in failing to accept the state appellate court’s factual determination that the prosecutor dismissed the five jurors, two of whom were not Latinos, because of their ability to speak Spanish and because the translation of taped conversations in Spanish was expected to be hotly contested at trial. We also hold that the Equal Protection Clause does not prohibit a trial attorney from peremptorily challenging jurors because of their ability to understand a foreign language the translation of which will be disputed at trial. We therefore reverse the orders of the district court.
I.
In April 1983, a New Jersey state court issued an order authorizing a wiretap of a telephone in Pemberthy’s apartment in Elizabeth. Conversations on that line were monitored and recorded for approximately one month. Most of these conversations were conducted in Spanish, and therefore they were monitored and translated into English by Spanish-speaking law enforcement officers. Law enforcement authorities concluded that many of the conversations, although *859cryptic, related to the importation of cocaine from Colombia.
In late May, Pemberthy and Moneada were arrested, and the police then executed search warrants for their residences and for various vehicles. In one of the vehicles, more than eight kilograms of cocaine were found. Pemberthy and Moneada, together with five others, were subsequently indicted by a state grand jury for three drug-related offenses: conspiracy to distribute and possess with intent to distribute one ounce or more of cocaine, simple possession of cocaine, and possession of cocaine with intent to distribute. Pemberthy was also indicted for an additional offense, theft of services.
Pemberthy and Moneada both moved to suppress all of the wiretap evidence, arguing, among other things, that the interceptions had not been properly minimized due to the monitors’ deficient knowledge of Spanish. As Moneada’s attorney later put it in his brief to the Appellate Division:
It is clear that part of the problem here came from the fact that non-professional Spanish-speaking “monitors” listened to these predominantly foreign conversations. The Detectives, Troopers and Officers apparently felt, in their own minds at least, that they had to listen to everything in order to get legitimate translation.
Moncada’s Br. at 44, State v. Moneada, 224 N.J.Super. 280, 540 A.2d 227 (App.Div.1988). Likewise, Pemberthy contended that “persons not sufficiently fluent in the language being spoken on the wiretaps could never minimize calls because they would have to listen over and over and for longer periods of time in order to get proper translations.” Pemberthy’s Br. at 9-10, State v. Pemberthy, 224 N.J.Super. 280, 540 A.2d 227 (App.Div.1988). See also id. at 17-19. After an extensive hearing on minimization and on the audibility of the tapes, the trial judge denied the motions to suppress and ruled that the taped conversations were admissible.
Jury selection took place on four days in June 1984, well before the New Jersey courts or the United States Supreme Court had held that peremptory challenges of individual jurors based on race or national origin violated either the state or federal constitution. See Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. 1712, 90 L.Ed.2d 69 (1986); State v. Gilmore, 103 N.J. 508, 511 A.2d 1150 (1986), aff'g 199 N.J.Super. 389, 489 A.2d 1175 (App.Div.1985). Because the translation of the tapes had been an issue at the suppression hearings and promised to be an issue at trial as well, the prosecutor asked the trial judge to question the prospective jurors about their ability to speak Spanish. The trial judge did so, and the prosecutor subsequently exercised the five peremptory challenges that are at issue here.
The first of these challenges concerned Alberto R. Casanova, who had been born in Cuba and had spoken Spanish all his life. J.A. 22, 45. In response to questioning by the court, Mr. Casanova stated that he spoke Spanish “perfectPy].” J.A. 42. The following exchange then occurred:
THE COURT: ... There will in this case, I know, be a lot of Spanish being spoken. ... [I]f you’re on the jury in this case, you’re going to hear some tapes of conversations, telephone calls, and they will be in Spanish predominantly....
Would you ... accept the interpretation as given with respect to the language ... and not substitute your own versions of what really it’s about, but accept what comes through from the Court?
Can you do that?
MR. CASANOVA: Certainly.
J.A. 42-43. The trial judge subsequently informed Mr. Casanova that not only would there be “a lot of Spanish-speaking witnesses,” but that not all would have the “same background.” J.A. 45. The court continued: “Some will be people that were born in Cuba, some may be from Puerto Rico, some may be officers who would be born here and speak Spanish.” J.A. 45. Mr. Casanova reiterated, however, that he could follow the court’s instruction. Id. After this colloquy, counsel for Pemberthy and Moneada stated that they would object to “every person who speaks Spanish being called to sidebar or grilled or questioned” (J.A. 46), and thereafter the trial judge did not question Spanish-speaking jurors in the same detail.
*860The next of the five jurors at issue was Gonzalo Quesada, who stated that he had spoken Spanish his entire life. J.A. 79. When the prosecutor asked the judge to inquire “where [Mr. Quesada] was from” (J.A. 84), the .following exchange occurred (J.A. 84-85):
THE . COURT: I'll ask the Spanish he knows.
[MONCADA’S COUNSEL]: Exactly.
THE COURT: Cuban. What is it? I want to know.
[MONCADA’S COUNSEL]: That was my question.
THE COURT: All right. Fine.
[End of side bar].
THE COURT: Mr. Quesada, sir, you mentioned you speak Spanish.
MR. QUESADA: Yes, sir.
THE COURT: All right. What is the Spanish you speak. What is it, Cuban Spanish?
MR. QUESADA: Cuban.
The third juror who spoke Spanish was Catherine Rocea, who stated that she had taught Spanish in high school. J.A. 159. Mrs. Rocca was not questioned about her ancestry.
The fourth Spanish-speaking juror was Irma Quinones, who stated that she was of Puerto Rican ancestry, that she had spoken Spanish since she was a young child, and that she now usually spoke Spanish only at work. J.A. 220-21. She also revealed that her nephew had been “caught breaking and entering” and that during the previous year her daughter, then 13, had been accused of attempting to murder an eight-year-old but that no charges had been brought. J.A. 217-19.
The final juror at issue was Mr. Bodet, who was originally from Haiti (J.A. 405) and stated that he “spoke [just a little bit] of Spanish” (J.A. 403). When informed that some of the defendants resided in Elizabeth, Mr. Bodet asked where they lived because he also resided in that city in an area with many Colombians. J.A. 400. It was discovered, however, that he did not live in the same part of the city. J.A. 401.
Defense counsel did not object to the prosecution’s exercise of peremptory challenges until after the jury had been empaneled and sworn. At that point, counsel for Pemberthy and Moneada moved for a mistrial. In addition to arguing that Hispanies had been underrepresented in the jury pool, counsel for Pemberthy argued that the prosecution had discriminated against blacks and Hispanies in the use of its peremptory challenges. Counsel for Pemberthy stated that “there’s no question in my mind that any juror that said they spoke Spanish was excused and that was more than just Hispanic jurors.” J.A. 441. When the court questioned whether every Spanish-speaking juror had in fact been stricken, Pemberth/s counsel responded:
Everyone. There was one juror that I remember specifically who was not His-panic who taught Spanish, and she was excused.
J.A. 441.
Immediately after defense counsel made their motion, the trial judge made the following statement (id. at 442-43):
By the way, let’s get something clear for the record. I think we all agree that we all wanted to know where the people did speak Spanish because that was something we agreed on, we wanted to find out that. In fact, I think it was a request put to me at the very outset....
Because of the fact that we’re going to have interpreters, and something on tape in Spanish, we at least wanted to know whether or not people might be second guessing the interpreters, or otherwise .... So, I agreed it was important to find out, so I asked all the jurors if they spoke Spanish, and several of them did.
The judge then asked the prosecutor whether he wanted to respond, and the prosecutor stated that he knew of no New Jersey case requiring a prosecutor to explain the basis for exercising individual peremptory challenges. J.A. 443. The judge agreed that that was the law “at least [at] this point in New Jersey,” and he added that he saw no systematic exclusion of Hispanies. J.A. 443-44. The judge observed that not all of the *861Spanish-speaking jurors had been “people with Hispanic backgrounds.” J.A. 444. The trial then began and lasted 20 days.
For present purposes, it is not necessary to review the evidence at trial. We note, however, that we agree with the following observation of the state Appellate Division: “Our review of the record satisfies us that the interpretation and translation of many of the telephone conversations constituted a major issue at the trial, and that many disputes centered on the interpretation of the statements made in the interpreted telephone conversations.” State v. Pemberthy, 224 N.J.Super. 280, 540 A.2d 227, 232 (App.Div.1988).
The prosecution introduced all of its translations of intercepted conversations through the testimony of New Jersey State Trooper Juan Cardenoza, and Cardenoza was extensively cross-examined regarding his proficiency in Spanish and his knowledge of Colombian colloquialisms and customs. See J.A. 1087-95. For example, Pemberthy’s lawyer asked Cardenoza: “Isn’t it a fact, investigator, that [Pemberthy] spoke a higher intelligence level of Spanish than you speak?” J.A. 1091. Cardenoza was also asked the following questions, among others:
Q. Isn’t it a fact that ... the type of Spanish spoken by Colombians is much more formal than the type of Spanish ... spoken by either Puerto Ricans or Cubans?
J.A. 1092.
Q. Isn’t it a fact, investigator, ... after getting used to speaking to Puerto Ri-cans and Cubans on the streets of New York, Elizabeth and Trenton, that when you talk to an educated Colombian everything sounds guarded?
J.A. 1094.
Q. ... What about street language? If they talk in street ... common street language would that make a difference to you in terms of the way you translated it? You know, street language in Colombia may be different than street language in Elizabeth. Does that make a difference to you?
J.A. 1089-90.
In their summations, counsel for both Pemberthy and Moneada emphasized the importance of the disputes regarding the translations. Counsel for Moneada stated: “I submit to you that there have been improper translations throughout the course of the transcript which have been read to you and I think we brought that' out on cross-examination.” 7/18/84 Tr. at 13. He later added:
[T]he point of the matter is that Trooper Cardenosa came to this country when he was only 16 years old. He doesn’t speak Spanish on a day-in-day-out basis 24 hours a day. He hasn’t done so for the last 16 years, okay? He lived in Colombia for a period of time. Is his Spanish.as good as Mr. Pemberthy’s or as good as Mr. Monca-da’s ...?
Id. at 26. He also discussed disputes about the translation of specific words and phrases. Id. at 12, 26, 43-44.
Counsel for Pemberthy returned to the same theme, stating with reference to Trooper Cardenoza:
The guy moved from Colombia when he was sixteen. He’s going to tell you that he knows what the Colombian expressions are and I’ll ask you to use your common sense and think about expressions that you used sixteen years ago and want you to tell me whether or not if you studied the English language whether the expressions that were in use in this country sixteen years ago are the same ones today. I doubt it....
Mr. Pemberthy has a degree and you!ll have it with you in the court room. He’s got a degree as an economist. Mr. Monea-da is the same and you heard Mr. Pember-thy talk through the translator.... There were times when his level of conversation was above even the court appointed translator, who’s trained to do it and then I’ll ask you whether or not Investigator Car-denosa who doesn’t have a degree from college or Mr. Diaz or Mr. Gonzalez [other monitors] whether they were even capable of understanding the level of conversation and Cardenosa admitted on the stand that *862that’s a difference because sometimes if you are not as polite and well spoken as the person’s whose words you are translating you’ll miss it....
Id. at 48, 50-51.
The jury found Pemberthy guilty on all counts. The jury also found Moneada guilty of conspiracy but not guilty of either simple possession or possession with intent to distribute. After sentencing, both defendants appealed.
By the time Pemberthy and Moneada filed their briefs with the Appellate Division, the United States Supreme Court had handed down its decision in Batson, and the New Jersey Supreme Court had issued its decision in Gilmore, which held that the state constitution imposes similar requirements. Accordingly, both Pemberthy and Moneada asserted, among many other arguments, that the prosecutor’s use of peremptory challenges was inconsistent with the holdings in those cases.
In a published opinion, the Appellate Division rejected these arguments.1 224 N.J.Super. 280, 540 A.2d 227. Under Gilmore, the Appellate Division observed, when a criminal defendant makes a prima facie showing that the prosecution exercised peremptory challenges on a constitutionally impermissible ground, the prosecution must “ ‘articulate “clear and reasonably specific” explanations of its “legitimate reasons” for exercising each of the peremptory challenges.’ ” 540 A.2d at 232 (quoting Gilmore, 511 A.2d at 1150). The Appellate Division did not address whether Pemberthy and Moneada had made a prima facie showing, but the court noted that the state had attempted to articulate a legitimate basis for its strikes, and the court accepted that explanation. The court wrote:
Here, the State concedes that Spanish-speaking jurors were excluded, but not on the basis of ethnicity. Thus, if a juror of Spanish ancestry who did not speak Spanish was on the panel, he would not have been excused. However, Spanish-speaking jurors without Spanish ancestry (such as the Spanish teacher) were excluded. The State explains that much of its evidence came from translations of Spanish conversations and questions of Colombian dialect were anticipated to potentially arise. Hence, the State wanted all the jurors to be operating on the same basis without having some jurors who thought they had greater knowledge of the Spanish usage and language than the translators who might be presented by the State and defendants. The substance of the Spanish telephone conversations was a serious issue during the trial. Defendants had the full opportunity to present their own translations and interpretations. Hence, this case does not fall within the ambit of Gilmore ....
We conclude that the State’s reason for exclusion was predicated on valid, articulated, trial-related reasons rather than on the basis of the potential juror’s presumed group bias.
540 A.2d at 233-34. The Appellate Division noted “the continuing criticism and problems that have arisen regarding peremptory challenges” (id. at 233), but concluded that “under the circumstances here, and as long as peremptory challenges are permissible, there was no reversible error” (id. at 234). Both Pemberthy and Moneada filed petitions for certification with the Supreme Court of New Jersey, but these petitions were denied. State v. Pemberthy, 111 N.J. 633, 546 A.2d 547 (1988); State v. Moneada, 111 N.J. 633, 546 A.2d 547 (1988).
Pemberthy and Moneada then filed petitions for writs of habeas corpus in the United States District Court for the District of New Jersey, arguing, among other things, that the prosecutor’s use of peremptory challenges violated their equal protection rights. After receiving briefs on this issue, the district court stated, it “realized that the prosecutor ... had never offered his reasons for the disputed challenges in the context of a Batson hearing.” Pemberthy v. Beyer, 800 *863F.Supp. 144, 150 (D.N.J.1992). The district court also observed that “[although the State Appellate Division implicitly accepted the Attorney General’s justification for the prosecutor’s use of peremptory challenges as presented on appeal, the Prosecuting Attorney himself did not proffer directly any justification for his peremptory strikes.” Id. at 151. The district court concluded that the Appellate Division had not made any factual findings concerning the reasons for the prosecutor’s challenges but had instead “focused on the legitimate reasons which might justify excluding Spanish-speaking jurors.” Id. at 152. The court thus decided that it should conduct a Batson hearing and proceeded to do so.
At the hearing, the prosecutor testified that he had stricken every Spanish-speaking juror, explaining that he had done so
[bjeeause I felt that the interpretation of the language would be a very integral part of the case. I didn’t want one or two people on the jury who might know Spanish who might disagree with the interpretation as counsel would put forth to have sort of like a more important place on a jury-
I wanted everybody on the same footing, to be able to judge the ease on the same footing. Didn’t want one juror to have a leg up on the other jurors because of special knowledge.
Id. at 153.
The prosecutor also offered supplemental reasons for striking several of the five jurors at issue. The prosecutor stated that he had dismissed Ms. Quinones in part because her daughter and nephew had been accused of committing serious crimes. Id. He stated that he had dismissed Mr. Bodet in part because it appeared that he might have known one of the defendants. Id. He also stated that he was concerned that Mr. Casanova might not accept the translations offered in court since he had stated that he spoke “perfect” Spanish. Id. at 154.
Following the hearing, the court granted the petitions. After finding that Pemberthy and Moneada had made out a prima facie case, the district court held that the state had failed to establish a legitimate justification for its peremptory challenges. The court based this conclusion on two alternative grounds. First, the court held that the state had not offered a legitimate race-neutral explanation for its peremptory challenges. Id. at 157-63. The court held, as we understand its opinion, that striking Latino jurors simply because they can speak Spanish is tantamount to striking them based on race. The court observed that “the one characteristic that links and, in essence, defines [the Latino] community/‘raee’ and exposes it to irrational discrimination is their status as native Spanish speakers.” Id. at 160 (emphasis in original). The court then reasoned: “Thus, the decision to strike all Spanish-speaking persons — especially native speakers who speak it ‘perfectly’ — is, by definition, a decision to strike all Latino jurors. In this sense, the prosecutor in this case ... offered a race-based reason for excluding jurors.” Id. The court further stated that the prosecutor’s explanation (i.e., that he thought that native Spanish speakers would have difficulty accepting translations that did not comport with their own understanding of conversations in Spanish) was based on “generalized, group-based assumptions about the capacity, behavior, and trustworthiness of Spanish speakers, without any individualized basis for concern.” Id. at 162.
The alternative ground for the district court’s decision was its finding that “the prosecutor in petitioner’s trial purposefully discriminated in the use of his peremptory strikes.” Id. at 164. The court found that “the prosecutor’s stated reason, i.e., concern about prospective jurors’ ability to accept the translations offered in court, was in fact a pretext for discrimination against members of the Latino community, who are defined and identified by their bilingual ability.” Id.
The district court entered orders granting Pemberthy’s and Moncada’s petitions but provided that its orders would be stayed if the state appealed. Id. at 168. The state then did so.
II.
We will first address the narrower of the two alternative grounds on which the district *864court based its decision, i.e., its finding that the prosecutor’s real reason for peremptorily challenging the five jurors in question was not concern about translation disputes that might arise during the trial but simple prejudice against Latinos. If this were a direct appeal from a federal criminal conviction, we would give substantial deference to the findings of the district court judge who presided over the jury selection process. See Hernandez v. New York, 500 U.S. 352, 364, 111 S.Ct. 1859, 1869, 114 L.Ed.2d. 395 (1991). This appeal, however, involves collateral attacks on state convictions, and thus both our court and the district court must defer to any factual determinations made by the state courts. Id. Under the federal habeas corpus statute, 28 U.S.C. § 2254(d)(8), if a state court of competent jurisdiction, after a hearing on the merits, makes a factual determination evidenced by “adequate written indicia,” the state court’s determination generally must be “presumed to be correct” unless it is not “fairly supported by the record.”2
The Supreme Court has held, moreover, that this provision applies, not only when a state trial court makes what are conventionally regarded as findings of fact, but also when a state appellate court makes factual determinations in a written opinion. Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981). In Sumner, a defendant in a state prosecution did not object at trial to his in-court identification by the state’s witness. Id. at 541, 101 S.Ct. at 766. On appeal from his conviction, he argued that the testimony of those witnesses was constitutionally tainted because of an unduly suggestive pretrial identification procedure, but the state intermediate appellate court’s opinion rejected this argument based on its factual determinations concerning the pretrial procedure. See id. at 542, 101 S.Ct. at 766. The Supreme Court in a subsequent federal habeas corpus proceeding held that the factual .determinations in the state appellate court’s' opinion were entitled to Section 2254(d)’s “ ‘presumption of correctness.’ ” Id. at 547, 101 S.Ct. at 769. See also, e.g., Hakeem v. Beyer, 990 F.2d 750, 768 (3d Cir.1993).
In this ease, the district court acknowledged that the “State Appellate Division implicitly accepted” the state’s contention that the prosecutor’s peremptory challenges had been based on the jurors’ ability to speak Spanish. 800 F.Supp. at 151. Nevertheless, the district court concluded that the Appellate Division made no “factual findings” on this issue. 800 F.Supp. at 151. We cannot agree with this interpretation of the Appellate Division’s decision.
From the Appellate Division’s opinion, it seems clear that that court made a factual determination that the prosecution exercised its peremptory challenges based on the ability of the jurors in question to speak Spanish. The Appellate Division first noted that it was the state’s position that the prosecutor had challenged these jurors because of their ability to speak Spanish and that this was a valid, trial-related reason. 540 A.2d at 233.3 The *865Appellate Division then accepted this argument, stating: “We conclude that the State’s reason for exclusion was predicated on valid, articulated, trial-related reasons rather than on the basis of the potential juror’s presumed group bias.” Id. at 234. These passages show that the Appellate Division made a factual determination regarding the prosecutor’s reason for exercising the peremptory challenges at issue in this appeal.
This conclusion is fortified by the fact that the Appellate Division could not have logically rejected Pemberthy’s and Monca-da’s Batson and Gilmore arguments without making such a determination. Batson held that a prosecutor may not exercise peremptory challenges based on race (476 U.S. at 89, 106 S.Ct. at 1719), and Gilmore made it clear that under the New Jersey Constitution this prohibition also extends to challenges based on “religious principles, ... color, ancestry, national origin, or sex.” 511 A.2d at 1158 (emphasis added). Pemberthy and Moneada argued that the prosecutor violated these precepts. State v. Pemberthy, 540 A.2d at 232. In response, the state “eoncede[d] that Spanish-speaking jurors were excluded,” but it argued that this exclusion was “not on the basis of ethnicity.” Id. at 233. Thus, as a simple matter of logic, the Appellate Division could not have rejected Pemberthy’s and Moncada’s argument without determining that the prosecutor’s challenges were not in fact based on ancestry or national origin. Accordingly, we are convinced that the Appellate Division found that the prosecutor’s peremptory challenges were based on Spanish-speaking ability and not on simple racial or ethnic prejudice.
This finding, furthermore, is “fairly supported by the record,” as 28 U.S.C. § 2254(d)(8) requires. First, it is apparent that the prosecutor in this ease had a legitimate basis for being concerned about whether the jurors ultimately chosen would accept the translations offered in court. As previously noted, the petitioners raised translation issues before trial; the trial judge questioned the prospective jurors about their ability to speak Spanish; and “the interpretation and translation of many of the telephone conversations constituted a major issue at the trial.” 540 A.2d at 232. The attorneys representing Pemberthy and Moneada vigorously challenged the ability of the state’s chief translator, Trooper Cardenoza, suggesting that his Spanish was rusty, that he was not capable of accurately translating the more “educated” level of Spanish supposedly spoken by their clients, and that he was not knowledgeable regarding Colombian customs, Colombian “street language,” and the latest Colombian colloquialisms.
Second, it is noteworthy that at the time when the jury selection in this case occurred and the prosecutor requested that the trial judge inquire during voir dire about the prospective jurors’ ability to speak Spanish, neither federal nor state case law forbade individual peremptory challenges based on race or national origin. This timing diminishes the likelihood that' the prosecutor made' his voir dire request for the purpose of constructing a plausible pretext for discrimination. In order to believe that this was the prosecutor’s purpose, one would have to believe that he foresaw the decisions in Batson and Gilmore and was already plotting to circumvent them.
Third, it is telling that two of the five jurors at issue — namely, Mrs. Rocca and Mr. Bodet — were not Latinos. As the Appellate Division found:
[T]he State exercised its peremptory challenges by excluding a Cuban, who spoke Spanish [Mr. Casanova]; a non-Hispanic who stated she taught high school Spanish [Mrs. Rocca];4 a Cuban who wrote and *866spoke Spanish [Mr. Quesada]; a native Puerto Rican who spoke Spanish, although mostly at work [Ms. Quinones]; and a Haitian who said that Colombians lived on his block and he spoke a little Spanish [Mr. Bodet].
540 A.2d at 232. The fact that the prosecutor struck all of the jurors who could speak Spanish, including two non-Latinos, certainly suggests that the prosecutor’s professed concern about translation issues was not a pretext.
Finally, if the briefs filed by Moneada and Pemberthy in the Appellate Division can be read as arguing that the prosecutor’s real reason for striking the Spanish-speaking jurors was prejudice against Latinos, the briefs advanced that argument obliquely and without marshaling any factual support. The heading of Point I of Moncada’s brief trumpeted the factual determination that the Appellate Division eventually made and that the district court rejected. That heading stated:
A NEW TRIAL IS MANDATED DUE TO THE STATE’S OBVIOUS PLAN TO SYSTEMATICALLY EXCLUDE ALL SPANISH-SPEAKING INDIVIDUALS FROM THE JURY PANEL.
Moncada’s Br. at 30 (emphasis added). Somewhat similarly, Pemberthy’s brief argued that “the State used its challenges to exclude anyone who even spoke Spanish.” Pemberthy’s Br. at 45. Although both briefs did assert in passing that the prosecutor had sought to exclude all “Spanish”5 or “hispanic”6 jurors, neither brief elaborated on or provided any factual support for this suggestion; nor did either brief attempt to reconcile this suggestion with the concession that the prosecutor had sought to dismiss all Spanish speakers. In light of this presentation and the facts that were evident from the record or were undisputed,7 it is entirely understandable why the Appellate Division made the factual determination that it did.
We have considered all of the factors cited by the district court in making a finding contrary to that of the Appellate Division,8 but these factors do not persuade us that the Appellate Division’s determination lacks fair support in the record. What seems to have figured most heavily in the district court’s analysis was the prosecutor’s professed belief that any Spanish-speaking juror might have difficulty accepting translations offered in court if those translations differed from the juror’s own understanding of conversations *867in Spanish.9 As the district court put it, “the prosecutor state[d] that no native speaker could ever set aside their knowledge of the language and restrict themselves to the evidence as presented,” and the district court viewed this attitude as persuasive evidence of bias. 800 F.Supp. at 165.
To test the strength of this analysis, .we think that the following mental experiment is instructive. Readers whose primary language is English should imagine that they are also proficient in another language and that they are serving on a jury in a jurisdiction in which that other language is spoken. Readers should also imagine that some of the evidence in the trial is in English, that this evidence is translated into the native language of the jurisdiction, and that the translation of some key passages seems, based on the readers’ knowledge of English, to be clearly wrong. Readers should then ask whether they could say without reservation that they could render a verdict based on the apparently erroneous translation provided in court — even if it seemed to them that a correct translation would dictate a contrary verdict. We suspect that many readers would find it difficult to say that under these circumstances they could unhesitatingly follow the translation offered in court. In any event, we do not think that it is strongly indicative of bias for a trial lawyer to suspect that some people, if placed in this situation, would find it difficult, at a conscious or subconscious level, to follow the translation.
This same point was made cogently in the brief that was filed in Hernandez v. New York, supra, by the Mexican-American Legal Defense and Educational Fund and the Commonwealth of Puerto Rico, Department of Puerto Rican Community Affairs in the United States. That brief noted that “[tjhere is a lay belief that bilinguals can turn off their Spanish language capability and receive information only in English” but that this cannot in fact be done. Br. at 5. .The brief contended that “[rjequiring bilingual Hispanics to allow to the fidelity of testimony translated into English adversely affects the majority of Hispanics by asking them to act in a manner they intuitively know is not possible.” Id. Asking a bilingual to “listen only to the English version of the testimony,” the brief maintained, is an “impossible ... task.” Id. at 13.10 For these reasons, we believe that the district court, in rejecting the prosecutor’s explanation for striking the five jurors, placed too much weight on the prosecutor’s assertion that any Spanish-speaking juror might have difficulty accepting the translations offered in court.
One other factor cited by the district court — what the court described as the prosecutor’s awareness of the “Latino ‘sympathy factor’ ” (800 F.Supp. at 164) — warrants discussion at this point. The district court wrote:
The prosecutor ... admitted that at the time of petitioner’s trial, he was aware of the “notion” that Latino jurors were potentially more sympathetic to Latino defendants- The prosecutor noted, howev•er, that he “didn’t think there would be a sympathy factor based upon ethnic background” because “most of my witnesses were also Spanish.”
800 F.Supp. at . 153 (citations omitted). The court then concluded that “the prosecutor’s admission that he was aware of the Latino ‘sympathy factor’” tended to suggest that the prosecutor “struck Latino jurors based in part on their shared race with the defen*868dants.” Id. at 164. In our view, however, this admission by the prosecutor could not reasonably be regarded as having much probative weight in the present context, for we doubt that there are many experienced trial lawyers who could truthfully say that they are not aware of the fact that some of their fellow practitioners have long believed — accurately or not11 — that jurors are more likely to be sympathetic to defendants of the same racial or ethnic background.12 It is thus difficult for us to see how this admission can be regarded as a significant indication that the prosecutor himself was prejudiced.13
In sum, we hold that the Appellate Division made a factual determination that the prosecutor peremptorily challenged the prospective jurors in question because of their ability to speak Spanish, that this determination is fairly supported by the record, and that it was therefore binding on the district court.
III.
In light of the Appellate Division’s factual determination, the remaining question that we must decide is whether the prosecutor violated the Equal Protection Clause by peremptorily challenging three Latinos and two non-Latinos because they could speak Spanish. At least with respect to the three Lati-. nos, this is a difficult question, as is evidenced by the various opinions in Hernandez. Justice Kennedy and the three other justices who joined his plurality opinion — the Chief Justice and Justices White and Souter — did not reach this question. Instead, they held that the prosecutor had not violated the Equal Protection Clause because he “did not rely on language ability without more, but explained that the specific responses and the demeanor of the two individuals during voir dire caused him to doubt their ability to defer to the official translation of Spanish-language testimony.” Hernandez, 500 U.S. at 360, 111 S.Ct. at 1867 (footnote omitted). Finding that this justification was permissible, Justice Kennedy’s opinion added:
We would face a quite different case if the prosecutor had justified his peremptory challenges with the explanation that he did not want Spanish-speaking jurors. It may well be, for certain ethnic groups and in some communities, that proficiency in a particular language, like skin color, should be treated as a surrogate for race under an equal protection analysis.
500 U.S. at 371, 111 S.Ct. at 1872-73.
The other two Justices in the majority did not join the plurality opinion. Asserting that the plurality opinion went “farther than it needs to” (id.), Justice O’Connor’s opinion, in which Justice Scalia joined, stated that “[n]o matter how closely tied or significantly correlated to race the explanation for a peremptory strike may be, the strike does not implicate the Equal Protection Clause unless it is based on race. That is the distinction between disproportionate effect, which is not sufficient to constitute an equal protection violation, and intentional discrimination, which is.” Id. at 375, 111 S.Ct. at 1874 (O’Connor, J., concurring in the judgment). It therefore appears that Justices O’Connor and Scalia would hold flatly that the Equal Protection Clause does not prohibit peremptory challenges that are sincerely based on jurors’ language ability. By contrast, the two dissenting Justices, Justices Stevens and Marshall, appear to have taken the position that the Equal Protection Clause forbids the striking of Spanish-speaking Latinos based *869on concern about their ability to accept translations offered in court. Id. at 375., Ill S.Ct. at 1875 (Stevens, J., dissenting).
In approaching the constitutional question that we must decide, we note at the outset that an equal protection violation cannot be established simply by showing that Latinos are disproportionately affected by peremptory challenges of jurors who can speak and understand Spanish. On this point, both the plurality and concurring opinions in Hernandez are in agreement. As the plurality wrote, “[e]qual protection analysis turns on the intended consequences of government classifications. Unless the government actor adopted a criterion with the intent of causing the impact asserted, that impact itself does not violate the principle of race-neutrality.” Id. at 362, 111 S.Ct. at 1867. See also id. at 372, 111 S.Ct. at 1873 (O’Connor, J., concurring in the judgment).
Moreover, it seems beyond reasonable dispute that there are at least some circumstances in which a peremptory challenge based on a juror’s ability to speak a foreign language would not violate the Equal Protection Clause. For example, suppose that the meaning of certain business documents written in a foreign language — say, German or Japanese — was a critical issue in a case and that by chance one of the prospective jurors was a professor, not of German or Japanese ancestry, who had learned the language in question while in college and graduate school. We find it hard to see how striking this juror could be regarded as tantamount to or as a proxy for racial or ethnic discrimination. Instead, striking this juror would appear to represent an application of the strategy, which is common among trial lawyers, of striking any prospective juror who might, as a result of education or employment, possess specialized information or expertise bearing on an important issue in the case.14 Based on this strategy, some trial lawyers are inclined to strike physicians and nurses in cases with medical issues, accountants in cases with accounting or. tax issues, engineers in cases with issues of product design, and so forth.15 Striking the hypothetical professor seems closely related to these examples, and we believe that it would be equally permissible under the Equal Protection Clause. Similarly, we do not think that striking a non-Latino juror based on that juror’s ability to speak Spanish — precisely what occurred here with respect to Mrs. Rocea and Mr. Bodet — raises a serious equal protection question.
We recognize, of course, that a more complex and debatable question is presented when a trial attorney strikes jurors based on their fluency in the language spoken in the country from which they or their ancestors emigrated. From the standpoint of sociology, the ability to speak a particular language may be an important facet of ethnic identity, both for the members of the ethnic group in question and for outsiders. As the Hernandez plurality noted, “[ljanguage permits an individual to express both a personal identity and membership in a community.” 500 U.S. at 370, 111 S.Ct. at 1872. Furthermore, as the Hernandez pluralityalso observed, individuals who continue to speak the language also observed, individuals. who continue to speak the language of the country to which they -trace their ancestry may, for that reason, be subjected by others to “ridicule,” “scorn,” and “hostility.” Id.
At the same time, however, no simple equation can be drawn between ethnicity and language. Sociologists recognize language as only one of the many components of ethniei*870ty.16 And while the district court in this case suggested that, for Latinos, ethnic identity is based exclusively on language, that view seems exaggerated. According to the 1990 census, approximately 25% of Hispanies do not claim proficiency in Spanish.17 In addition, there are many non-Hispanics, like Mrs. Rocca and Mr. Bodet, who have acquired some proficiency in Spanish through education, employment, travel, or other experiences.
In any event, the question that faces us in this case is not the degree to which ethnicity and language are linked from the perspective of sociology. Rather, the question is whether peremptory challenges based on language ability are equivalent for equal protection purposes to the types of challenges prohibited in Batson and related cases. We hold that they are not.
Ethnicity is an immensely complex phenomenon. All the groups treated here are characterized by some of the following features, although in combinations that vary considerably:
(1) common geographic origin;
(2) migratory status;
(3) race;
(4) language or dialect;
(5) religious faith or faiths;
(6) ties that transcend kinship, neighborhood, and community boundaries;
(7) shared traditions, values, and symbols;
(8) literature, folklore, and music;
(9) food preferences;
(10) settlement and employment patterns;
(11) special interests in regard to politics in the homeland and in the United States;
(12) institutions that specifically serve and maintain the group;
(13) an internal sense of distinctiveness;
(14) an external perception of distinctiveness.
We believe that Batson does not apply to ■ peremptory challenges unless they are based on classifications, such as race or national origin, that are subject to “strict” scrutiny under equal protection doctrine, or possibly those classifications, such as gender, that are subjected to “heightened” scrutiny.18 Classifications based on the ability to speak or understand a foreign language do not meet the requirements for either “strict” or “heightened” scrutiny. As the Supreme Court explained in Cleburne v. Cleburne Living Center, Inc., 473 U.S. 432, 440, 105 S.Ct. 3249, 3254, 87 L.Ed.2d 313 (1985), one of the important reasons why classifications based on race and national origin are subjected to strict scrutiny is because these characteristics are “so seldom relevant to the achievement of any ■ legitimate state interest that laws grounded in such considerations are deemed to reflect prejudice and antipathy — a *871view that those in the burdened class are not as worthy or deserving as others.” The Court went on to explain that gender classifications are subjected to “heightened” review in large measure because gender “generally provides no sensible ground for differential treatment.” Id. By contrast, the Court observed that classifications based on age are not subjected to heightened review, in part because persons in different age groups have “distinguishing characteristics relevant to interests the State has the authority to implement” and because the aged have not been subjected to a history of intentional unequal treatment. Id. at 441, 105 S.Ct. at 3255. Relying on similar reasoning, the Court then held that classifications based on mental retardation should not be subject to a heightened standard of review. Id. at 442, 105 S.Ct. at 3255.
In light of this analysis, we do not think that a classification based on the ability to speak a foreign language can qualify' for “strict” or “heightened” review. This ability may in many instances be “relevant to the achievement of [a] legitimate state interest.” 473 U.S. at 440, 105 S.Ct. at 3254.19 Furthermore, linguistic ability is not immutable, and the history of discrimination against individuals who speak a foreign language in addition to English is not comparable to the history of discrimination based on factors such as race or national origin.
For these reasons, we are not willing to hold as a matter of law that language-based classifications are always a proxy for race or ethnicity and receive strict scrutiny for that reason; nor are we willing to, hold that language-based classifications receive strict or heightened scrutiny for any other reason. However, we wish to emphasize in the strongest terms that a challenger’s decision that is actually motivated by racial or ethnic considerations continues, to be subject to strict scrutiny even when the attorney asserts that he or she is categorizing jurors by *872linguistic ability rather than by race or ethnicity.
Under this rule, trial attorneys are not free to strike Latino jurors in every case that features some testimony or evidence in Spanish. Just because an attorney has a theoretically rational reason for striking Spanish-speaking jurors does not mean that that is the attorney’s actual motivation. The attorney may be attempting to eliminate Latinos from the jury because of his or her view about how Latinos are likely to see a case. Moreover, an attorney cannot strike a juror because the attorney believes that it is “rational” in a particular ease to eliminate jurors based on ethnicity. Just as a challenger cannot escape strict scrutiny by showing that a racial or ethnic classification is rational, a challenger cannot escape strict scrutiny by arguing that a language-based classification that is a proxy for ethnicity is rational. And just as a prosecutor cannot strike Latino jurors based on a statistical study showing that Latino jurors are more likely to be sympathetic to X type of defendants than an average juror, a prosecutor cannot strike Spanish-speaking jurors based on a statistical study showing that Spanish-speaking jurors are more likely to be sympathetic to X type of defendants. Such a “rational” generalization about Spanish-speaking jurors would have little to do with the fact that the jurors spoke Spanish and a lot to do with the fact that most Spanish-speaking jurors are Latino. Thus, it would be the same as an impermissible generalization about Latinos. To reiterate, a challenger’s decision to strike jurors based on language ability is subject to rational basis review if and only if the challenger’s concerns have to do with language rather than ethnicity. The dispositive question is the factual question of subjective intent.
Because language-speaking ability is so closely correlated with ethnicity, a trial court must carefully assess the challenger’s actual motivation even where the challenger asserts a rational reason to discriminate based on language skills. In assessing that motivation, the trial court should consider among other factors 1) any extrinsic evidence of motivation, 2) whether the prosecutor’s strikes correlate better with language ability or with race, and 8) how strong the challenger’s reasons are to fear that translation issues will present a problem. (If the circumstances are such that a reasonable attorney would not be concerned about translation problems, the trial judge should be more suspicious that the attorney’s motivation is illicit.) Here, for example, the state court’s finding that the prosecutor was not motivated by race was reasonable because 1) this case was pre-Batson; 2) the translation issue was a central one in the case; and 3) the prosecutor also struck non-Latino, Spanish-speaking venirepersons. See supra, pp. 865-68. However, the state court might also have been'within its discretion to come to the opposite result.
Batson’s scheme for preventing discrimination in the exercise of peremptory challenges is founded on the belief that trial judges are generally capable of detecting instances in which attorneys’ professed reasons for peremptory challenges are pretextual, even if subtle and difficult determinations must be made. For example, under Batson, trial judges are trusted to make determinations such as whether a peremptory challenge was sincerely based on a juror’s failure to make eye contact20 or a juror’s facial expressions,21 “body language,”22 or fidgeting.23 If trial judges are able to make determinations such as these, we see no reason to believe that they are any less able to detect instances in which professed concerns about translation issues are pretextual. Nor do we have any doubt that both the federal and state trial *873judges in this circuit will make a sincere and vigilant effort to prevent discrimination against Latino jurors. As part of that effort, trial judges should be particularly sensitive to the potential use of language-based per-emptories for discriminatory purposes.
If these efforts by trial judges prove to be insufficient, then the federal and state statutes or rules concerning peremptory challenges can and should be amended. It would be utterly unacceptable if Latinos were commonly excluded from juries based on pretex-tual concerns about translations. Our task in this case, however, is limited to an analysis and application of the requirements imposed by the Equal Protection Clause, and under that Clause, for the reasons we have explained, peremptory challenges that are sincerely based on translation concerns are not prohibited.
IY.
For the reasons explained above, the orders of the district court are reversed.
SUR PETITION FOR REHEARING
May 6, 1994.
Present: SLOVITER, Chief Judge, and BECKER, STAPLETON, MANSMANN, GREENBERG, HUTCHINSON, SCIRICA, COWEN, NYGAARD, ALITO, ROTH and LEWIS, Circuit Judges.The petition for rehearing filed by appel-lees in the above-entitled case having been submitted to the judges who participated in the decision of this court and to all the other available circuit judges of the circuit in regular active service, and no judge who concurred in the decision having asked for rehearing, and a majority of the circuit judges of the circuit in regular active service not having voted for rehearing by the Court in banc, the petition for rehearing is denied.
. The court noted the State’s argument that Pem-berthy's and Moncada’s objections to the peremptory challenges were untimely under state law, but the court stated that it "prefer[red] to deal with this issue on the merits" because Gilmore had not been decided at the time of jury selection. 540 A.2d at 233 n. 6.
. The exact language of 28 U.S.C. § 2254(d)(8) is as follows:
In any proceeding instituted in a Federal court by an application for a writ of habeas corpus by a person in custody pursuant to the judgment of a State court, a determination after a hearing on the merits of a factual issue, made by a State court of competent jurisdiction in a proceeding to which the applicant for the writ and the State or an officer or agent thereof were parties, evidenced by a written finding, written opinion, or other reliable and adequate written indicia, shall be presumed to be correct, ...
(8) ... unless that part of the record of the State court proceeding in which the determination of such factual issue was made, pertinent to a determination of the sufficiency of the evidence to support such factual determination, is produced as provided for hereinafter, and the Federal court on a consideration of such part of the record as a whole concludes that such factual determination is not fairly supported by the record....
. The court wrote:
Here, the State concedes that Spanish-speaking jurors were excluded, but not on the basis of ethnicity.... The State explains that much of its evidence came from translations of Spanish conversations and questions of Colombian dialect were anticipated to potentially arise. Hence, the State wanted all the jurors to be operating on the same basis without having some jurors who thought they had greater knowledge of the Spanish usage and language than the translators who might be presented by the State and defendants.
. It is apparent that the Appellate Division made a factual determination, within the meaning of 28 U.S.C. § 2254(d), that Mrs. Rocca was not a Latino. In the excerpt from its opinion quoted above, the Appellate Division referred to her as "a non-Hispanic who stated she taught high school Spanish.” 540 A.2d at 232. The Appellate Division later stated that the prosecutor had dismissed "Spanish-speaking jurors without Spanish ancestry (such as the Spanish teacher).” Id. at 233. This finding is fairly supported by the record. As previously noted, counsel for one of the petitioners referred during the jury selection process to "one juror ... who was not Hispanic who taught Spanish.” J.A. 441. Moreover, both counsel for Pemberthy and the trial court noted that not all of the Spanish-speaking jurors had been Hispanic. See J.A. 441, 444. Furthermore, *866although the state’s Appellate Division brief argued, citing this portion of the record, that Mrs. Rocca was not Hispanic (state’s Moneada br. at 19; state's Pemberthy br. at 54 n. 22), the briefs filed by Pemberthy and Moneada did not argue that Mrs. Rocca was or might be a Latino. Thus, the district court erred in concluding that Mrs. Rocca's ethnicity "remains unknown.” 800 F.Supp. at 148.
. Moncada's brief (at 30) stated that "the State used many of its peremptory challenges in an attempt to exclude all Spanish and Spanish-speaking individuals from the panel.”
. Pemberthy’s brief (at 45) stated that "the State exercised [its] peremptory challenges to exclude all hispanic jurors merely because they were hispanic.”
. These include the importance of translation disputes at trial, the fact that the prosecutor had sought to exclude all Spanish-speaking individuals, and the fact that two of the five Spanish-speaking jurors were not Latinos.
. The district court based its finding on the following seven factors:
(1) the prosecutor's stated intention to exclude all Spanish speakers strikes dangerously close to the defining racial and/or ethnic characteristic of Latinos; (2) no individual response or behavior could dissuade the prosecutor from his decision to strike all Spanish speakers; (3) the prosecutor gave no consideration to any alternative or more accurate means to satisfy his concerns with potential jurors; (4) the prosecutor’s reason to suspect the abilities and anticipated behavior of Spanish speakers is premised on unfounded discriminatory and prejudiced assumptions about bilingual Spanish speakers, especially native speakers; (5) the prosecutor's attempt to provide an individualized, race-neutral explanation for striking jurors Casanova and Bodet is not credible; (6) the prosecutor’s admission that he was aware of the Latino "sympathy factor,” coupled with all the other aforementioned prejudices, suggests that the prosecutor struck Latino jurors based in part on their shared race with the defendants; and (7) the clearly foreseeable effect of the prosecutor’s system for striking jurors suggests that the prosecutor purposefully intended to exclude Latinos from the jury.
800 F.Supp. at 164.
. At least three of the seven factors cited by the district court (factors (2), (3), and (4)) appear to have been based entirely on this aspect of the prosecutor's thinking, and one additional factor (factor (5)) appears to have been based in part on the same ground. In factor (5), the district court found that the prosecutor's explanation for striking Mr. Casanova was not credible. The prosecutor explained that he had dismissed Mr. Casanova in part because Mr. Casanova stated that his Spanish was "perfect” and the prosecutor therefore questioned • whether Mr. Casanova could “put aside his knowledge of Spanish” and accept the translations offered in court. 800 F.2d at 154. The district court concluded that this attitude reflected "prejudiced assumptions about bilingual Latinos,” viz., that bilingual Latinos may find it difficult to accept an English translation as opposed to their own understanding of testimony or evidence in Spanish. Id. at 165.
. Commentators have made the same argument. See, e.g., Deborah A. Ramirez, Excluded Voices: The Disenfranchisement of Ethnic Groups From Jury Service, 1993 Wisc.L.Rev. 761, 777-85 (1993); Andrew McGuire, Peremptory Exclusion of Spanish-Speaking Jurors, Could Hernandez v. New York Happen Here?, 23 N.M.L.Rev. 467, 472 (1993).
. See Barbara D. Underwood, Ending Race Discrimination in Jury Selection: Whose Right Is It, Anyway? 92 Col.L.Rev. 725, 731-32 (1992) (summarizing social science research).
. Indeed, trial manuals advise attorneys to consider this factor in selecting juries. See, e.g., Herbert A. Kuvin, Trial Handbook 91-92 (1965).
. While we will not discuss in detail the remaining factors cited by the district court (see footnote 8, supra), we do not believe that they undermine the validity of the Appellate Division's factual determination. Factors (1) and (7) essentially embody the proposition that peremptorily challenging all Spanish speakers has a disproportionate impact on Latinos. Such an impact may be, but of course is not always, evidence of a discriminatory purpose or intent. See Hernandez, 500 U.S. at 360, 111 S.Ct. at 1867. The only other factor, factor (5), is based on the district court's finding that the prosecutor’s explanations of his supplemental reasons for striking jurors Casanova and Bodet were not credible. Insofar as this factor relates to Mr. Casanova, it has already been discussed. See footnote 9, supra. As far as Mr. Bodet is concerned, we note that he is not a Latino. See page 860, supra.
. A popular trial manual explains:
Care should be exercised in the selection of the jury to prevent the submission of the issues to a so-called "one-man jury.” A juror who can claim expertise in an important aspect of the case has an inordinate amount of influence on fellow jurors who are likely to respect his authority. Initially it might seem preferable to have someonfe on the jury who fully understands the issues, but there is a real danger that the "expert” juror will disagree with the attorney's theory of the case or the opinions of his experts. The juror’s experience, reading or school of thought may lead him to a conclusion contrary to the one which the trial attorney is advancing. The other jurors, who are aware of his superior knowledge and experience in the particular field, will naturally ask him for his opinions and ideas. In all probability, assuming the friendships usually created during jury service, the jurors will go along with the "expert" juror's judgment. In this manner, it is possible for one person to decide the case. Submitting a case to a "one-man juiy” is an enormous risk.
1 Fred Lane, Lane's Goldstein Trial Technique § 9.54 (3d ed.1984).
. See id. at §§ 9.55, 9.79.
. For example, the Harvard Encyclopedia of American Ethnic Groups vi (Stephan Themstrom, et al., eds., 1980) states:
. Census Bureau, U.S. Dep't of Commerce, 1990 CP-1-1, 1990 U.S. Census of Population: General Population Characteristics, United States 3 tbl. 3 (Nov. 1992).
. Under familiar equal protection doctrine, the level of scrutiny that a court must employ depends on the nature of the classification at issue. "Classifications based on race or national origin ... and classifications affecting fundamental rights ... are given the most exacting scrutiny.” Clark v. Jeter, 486 U.S. 456, 461, 108 S.Ct. 1910, 1914, 100 L.Ed.2d 465 (1988) (citations omitted). Such classifications must be narrowly tailored to serve a “compelling state interest.” Cleburne v. Cleburne Living Center, Inc., 473 U.S. 432, 440, 105 S.Ct 3249, 3254, 87 L.Ed.2d 313 (1985). Classifications based on sex or illegitimacy are subject to a level of "intermediate" or "heightened” scrutiny. Clark, 486 U.S. at 461, 108 S.Ct. at 1914; Cleburne, 473 U.S. at 440-41, 105 S.Ct. at 3255-55. This scrutiny requires a showing that “the classification served ‘important governmental objectives and that the discriminatory means employed’ are 'substantially related to the achievement of those objectives.' " Mississippi University for Women v. Hogan, 458 U.S. 718, 724, 102 S.Ct. 3331, 3336, 73 L.Ed.2d 1090 (1982) (quoting Wengler v. Druggists Mutual Ins. Co., 446 U.S. 142, 150, 100 S.Ct. 1540, 1545, 64 L.Ed.2d 107 (1980)). Finally, classifications based on almost all other social and economic factors are given the minimum level of scrutiny, which demands only that the classification be ratipnally related to a legitimate governmental purpose. See, e.g., New Orleans v. Dukes, 427 U.S. 297, 303, 96 S.Ct. 2513, 2516, 49 L.Ed.2d 511 (1976); San Antonio School District v. Rodriguez, 411 U.S. 1, 17, 93 S.Ct. 1278, 1288, 36 L.Ed.2d 16 (1973).
The constitutionality of peremptory challenges, in our view, must be judged using these same levels of scrutiny. Challenges based on the suspect factors of race or national origin are thus subject to strict scrutiny and, as Batson and related cases demonstrate, they cannot survive this level of review. Such peremptories are not narrowly tailored to serve the presumably compelling objective of obtaining a fair and impartial juiy; and although such peremptories may in some instances increase litigants' "acceptance of a jury panel as fair'.’ (Edmonson v. Leesville Concrete Co., Inc., 500 U.S. 614, 630, 111 S.Ct. 2077, 2088, 114 L.Ed.2d 660 (1991)), that interest is not compelling.
Just as challenges based on race or national origin are subject to strict scrutiny, it has been *871argued that challenges based on gender should be governed by "intermediate” or "heightened" scrutiny. See United States v. De Gross, 960 F.2d 1433, 1439 (9th Cir.1992). The Supreme Court is currently considering whether such challenges are prohibited by the Equal Protection Clause. See J.E.B. v. T.B., No. 92-1239 (argued 11/2/93).
Most other peremptory challenges, including those based on such common grounds as occupation and age, are presumably subject to only the minimum level of equal protection scrutiny. Thus, it seems likely that challenges of this sort, if tested, would be held to be- at least rationally related to the legitimate objectives of promoting both jury impartiality and the appearance of impartiality. See, e.g. United States v. De Gross, 960 F.2d at 1438 n. 8.
. This conclusion is quite obvious when a government actor favors individuals who are able to speak a foreign language. Suppose, for example, that a state or municipal agency is hiring employees for positions requiring frequent communications with members of a minority community in which a foreign language is commonly spoken and that the agency has a policy of favoring applicants who can speak that language. Or suppose that a state college or university gives a preference in admissions to students who have mastered two languages. We do not believe that such policies should be subjected to strict or heightened scrutiny.
It may, of course, be argued that classifications disfavoring or disqualifying those who can speak another language should not stand on the same footing as classifications favoring members of the same group, but the Supreme Court has repeatedly held that the appropriate level of equal protection scrutiny does not vary depending on whether a burden or benefit is being conferred. For example, the Court has held that all racial classifications require strict scrutiny regardless of whether they favor or disfavor the members of any particular group. See Richmond v. J.A. Croson Co., 488 U.S. 469, 494, 109 S.Ct. 706, 722, 102 L.Ed.2d 854 (1989) (plurality); id. at 520, 109 S.Ct. at 726 (Scalia, J., concurring in the judgment); Wygant v. Jackson Board of Education, 476 U.S. 267, 273, 106 S.Ct. 1842, 1846, 90 L.Ed.2d 260 (1986) (plurality); University of California Regents v. Bakke, 438 U.S. 265, 291, 98 S.Ct. 2733, 2748, 57 L.Ed.2d 750 (1978) (opinion of Powell, J.). The Court has likewise held that gender classifications are always subject to the same level of scrutiny no matter which sex is favored or disfavored. Mississippi University for Women, 458 U.S. at 723, 102 S.Ct. at 3335; Caban v. Mohammed, 441 U.S. 380, 394, 99 S.Ct. 1760, 1769, 60 L.Ed.2d 297 (1979); Orr v. Orr, 440 U.S. 268, 279, 99 S.Ct. 1102, 1111, 59 L.Ed.2d 306 (1979). It must follow, therefore, that all classifications based on the ability to speak or understand a foreign language are subject to the same level of equal protection scrutiny, namely, rational-basis review.
Of course, in the vast majority of circumstances, the ability to speak a second language “cannot reasonably be regarded as harmful.” Hernandez, 500 U.S. at 371, 111 S.Ct. at 1872, quoting Meyer v. Nebraska, 262 U.S. 390, 400, 43 S.Ct. 625, 627, 67 L.Ed. 1042 (1923). Consequently, such classifications may often fail to satisfy even the lowest level of equal protection scrutiny. .In addition, state laws or policies discriminating against individuals who are able to speak a foreign language may be challenged under other constitutional provisions, as Meyer v. *872Nebraska, supra, a substantive due process case, suggests.
. See, e.g., United States v. Casper, 956 F.2d 416, 418-19 (3d Cir.1992); United States v. Lance, 853 F.2d 1177, 1181 (5th Cir.1988).
. See, e.g., Reynolds v. Benefield, 931 F.2d 506, 512 (8th Cir.), cert. denied, 501 U.S. 1204, 111 S.Ct. 2795, 115 L.Ed.2d 969 (1991); Barfield v. Orange County, 911 F.2d 644, 648 (11th Cir.1990), cert. denied, 500 U.S. 954, 111 S.Ct. 2263, 114 L.Ed.2d 715 (1991).
. See, e.g., United States v. Johnson, 4 F.3d 904, 913 (10th Cir.1993), cert. denied, — U.S. -, 114 S.Ct. 1082, 127 L.Ed.2d 348 (1994).
. See, e.g., United States v. Power, 881 F.2d 733, 740 (9th Cir.1989).