Plaintiffs, Mexican-American children and their parents who represent a class of others similarly situated, instituted this action against the Raymondville, Texas Independent School District (RISD) alleging that the district engaged in policies and practices of racial discrimination against Mexican-Americans which deprived the plaintiffs and their class of rights secured to them by the fourteenth amendment and 42 U.S.C. § 1983 (1976), Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d et seq. (1976), and the Equal Educational Opportunities Act of 1974, 20 U.S.C. § 1701 et seq. (1976). Specifically, plaintiffs charged that the school district unlawfully discriminated against them by using an ability grouping system for classroom assignments which was based on racially and ethnically discriminatory criteria and resulted in impermissible classroom segregation, by discriminating against Mexican-Americans in the hiring and promotion of faculty and administrators, and by failing to implement adequate bilingual education to overcome the linguistic barriers that impede the plaintiffs’ equal participation in the educational program of the district.1 The original complaint also named the Secretary of the Department of Health, Education and Welfare (HEW) as a defendant and alleged that the department, although charged with responsibility to assure that federal funds are spent in a nondiscriminatory manner and cognizant of the school district’s noncompliance with federal law, had failed to take appropriate action to remedy the unlawful practices of the school district or to terminate its receipt of federal funds. By an amended complaint, the plaintiffs also named the Texas Education Agency (TEA) as a defendant and charged that the TEA had failed to fulfill its duty to assure that the class represented by the plaintiffs was not subjected to discriminatory practices through the use of state or federal funds.
The case was tried in June 1978; on August 17, 1978 the district court entered judgment in favor of the defendants based upon its determination that the policies and practices of the RISD, in the areas of hiring and promotion of faculty and administrators, ability grouping of students, and bilingual education did not violate any constitutional or statutory rights of the plaintiff class. From that judgment, the plaintiffs have brought this appeal in which they claim the district court erred in numerous matters of fact and law.
Although upon motion of the plaintiffs, HEW was dismissed as a defendant in this suit before trial, the agency remains an important actor in our current inquiry because this private litigation involves many of the same issues considered in an HEW administrative investigation and fund termination proceeding involving RISD. In April 1973, following a visit from representatives of HEW’s Office for Civil Rights (OCR), HEW notified RISD that it failed to comply with the provisions of Title VI and administrative regulations issued by the Department to implement Title VI. HEW requested that RISD submit an affirmative plan for remedying these deficiencies. Ap*993parently, RISD and the OCR were unable to negotiate a mutually acceptable plan for compliance and in June 1976, formal administrative enforcement proceedings were instituted in which the OCR sought to terminate federal funding to RISD. RISD requested a hearing on the allegations of noncompliance and in January 1977, a five day hearing was held before an administrative law judge. Thereafter, the judge entered a decision which concluded that RISD was not in violation of Title VI or the administrative regulations and policies issued thereunder. The judge ordered that the suspension of federal funds to the district be lifted. This decision was affirmed in April 1980, by a final decision of the Reviewing Authority of the OCR.
The extensive record of these administrative proceedings, including the transcript of the hearing before the administrative law judge and the judge’s decision, was received into the record as evidence in the trial of this case and included in the record on appeal. The defendants have moved to supplement the appellate record by including the decision of the Reviewing Authority. This motion was carried with the appeal. Since the record in this case already includes extensive material from this administrative proceeding, which involved many of the same questions of fact and law as this case, we see no reason why the final administrative determination of those questions should not also be included. The defendants’ motion to supplement the appellate record in this cause to include the final decision of the Reviewing Authority of OCR is, therefore, granted.
Before we turn to consider the specific factual and legal issues raised by the plaintiffs in their appeal of the district court’s judgment, we think it helpful to outline some of the basic demographic characteristics of the Raymondville school district. Raymondville is located in Willacy County, Texas. Willacy County is in the Rio Grande Valley; by conservative estimate based on census data, 77% of the population of the county is Mexican-American and almost all of the remaining 23% is “Anglo.” The student population of RISD is about 85% Mexican-American.
Willacy County ranks 248th out of the 254 Texas counties in average family income. Approximately one-third of the population of Raymondville is composed of migrant farm workers. Three-quarters of the students in the Raymondville schools qualify for the federally funded free school lunch program. The district’s assessed property valuation places it among the lowest ten percent of all Texas counties in its per capita student expenditures.
The district operates five schools. Two campuses, L.C. Smith and Pittman, house students in kindergarten through fifth grade. The student body at L.C. Smith is virtually 100% Mexican-American; Pittman, which has almost twice as many students, has approximately 83% Mexican-American students. There is one junior high school, which has 87% Mexican-American students, and one high school, in which the enrollment is 80% Mexican-American.
I. A THRESHOLD OBSTACLE TO APPELLATE REVIEW
In their brief on appeal, the plaintiffs contend first, that the analysis of the memorandum opinion in which the district court concluded that the challenged policies and practices of the RISD did not violate the fourteenth amendment, Title VI or the Equal Educational Opportunities Act is pervasively flawed by the court’s failure to make findings concerning the history of discrimination in the RISD in assessing the plaintiffs’ challenges to certain current policies and practices. Plaintiffs contend that these issues were properly raised by the pleadings and that there was ample evidence in the record to support findings that RISD had, in the past, segregated and discriminated against Mexican-American students and that, as yet, RISD has failed to establish a unitary system in which all vestiges of this earlier unlawful segregation have been eliminated because the virtually 100% Mexican-American school, L.C. Smith, is a product of this earlier unlawful policy of segregation. Although the plaintiffs in this case did not challenge the current stu*994dent assignment practices of the RISD (which are no longer based on attendance zones but rather on a freedom of choice plan) or request relief designed to alter the ethnic composition of the student body at L.C. Smith, the evidence of past segregative practices of RISD was relevant to the legal analysis of two of the claims the plaintiffs did make.
The plaintiffs here challenge the RISD’s ability grouping system which is used to place students in particular sections or classes within their grade. We have consistently stated that ability grouping is not per se unconstitutional. In considering the propriety of ability grouping in a system having a history of unlawful segregation, however, we have cautioned that if testing or other ability grouping practices have a markedly disparate impact on students of different races and a significant racially segregative effect, such practices cannot be employed until a school system has achieved unitary status and maintained a unitary school system for a sufficient period of time that the handicaps which past segregative practices may have inflicted on minority students and which may adversely affect their performance have been erased. United States v. Gadsden County School District, 572 F.2d 1049 (5th Cir. 1978); Morales v. Shannon, 516 F.2d 411 (5th Cir. 1975); McNeal v. Tate County School District, 508 F.2d 1017 (5th Cir. 1975); Moses v. Washington Parish School Board, 456 F.2d 1285 (5th Cir. 1972); Lemon v. Bossier Parish School Board, 444 F.2d 1400 (5th Cir. 1971); Singleton v. Jackson Municipal Separate School District, 419 F.2d 1211, 1219 (5th Cir. 1969).
The question whether RISD has a history of unlawful discrimination is also relevant to the analysis of plaintiffs’ claim regarding the district’s employment practices. In cases involving claims similar to those made here regarding a pattern or practice of discrimination in the employment of faculty and staff, we have held that when such a claim is asserted against a school district having a relatively recent history of discrimination, the burden placed on the defendant school board to rebut a plaintiff’s prima facie case is heavier than the burden of rubuttal in the usual employment discrimination case. In a case involving a school district with a history of discrimination, the defendant must rebut the plaintiff’s prima facie case by clear and convincing evidence that the challenged employment decisions were motivated by legitimate nondiscriminatory reasons. Lee v. Conecuh County Board of Education, 634 F.2d 959 (5th Cir. 1981); Lee v. Washington County Board of Education, 625 F.2d 1235, 1237 (5th Cir. 1980); Davis v. Board of School Commissioners, 600 F.2d 470, 473 (5th Cir. 1979); Hereford v. Huntsville Board of Education, 574 F.2d 268, 270 (5th Cir. 1978); Barnes v. Jones County School District, 544 F.2d 804, 807 (5th Cir. 1977). This, of course, is a much heavier burden of rebuttal than that imposed on an employer in the usual employment discrimination case under Texas Department of Community Affairs v. Burdine,-U.S.-,-, 101 S.Ct. 1089, 1094, 67 L.Ed.2d 207 (1981).2
Plaintiffs raised the issue of RISD’s past discrimination in their pleadings and introduced substantial evidence in support *995of this claim in the proceedings before the district court;3 thus, the district court’s failure to make findings regarding the history of the district and whether vestiges of past discrimination currently exist in the district cannot be excused on the grounds that these issues were not properly before the court. The absence of findings on these issues seriously handicaps our review of the merits of the ability grouping and employment discrimination claims made by the plaintiffs in this case. With regard to plaintiffs’ first two arguments on appeal, our opinion will, therefore, be limited to identifying the factual and legal determinations which, although necessary to a proper analysis of the plaintiffs’ claims, were not made by the district court and must be *996made upon remand and to reviewing those aspects of the merits of these claims which are not affected by this failure to make certain essential findings.
II. ABILITY GROUPING
RISD employs an ability grouping system of student assignment. In the elementary grades and the junior high school, students are placed in a particular ability group (labeled “high,” “average” or “low”) based on achievement test scores, school grades, teacher evaluations and the recommendation of school counselors. In grades 1-6, once students have been placed in a particular ability group, they are assigned to a specific class for that group by a random manual sorting system designed to assure that each classroom has a roughly equal number of girls and boys. After the junior high school students are grouped by ability, they are assigned to particular sections of their ability group by computer. Although Raymondville High School offers courses of varying pace and difficulty, students are not assigned to particular ability groups. High school students, with the assistance of their parents and school counselors, choose the subjects they wish to study (subject, of course, to the usual sort of prerequisites and curriculum required for graduation) and are free to select an accelerated, average or slower class. Plaintiffs claim that these ability grouping practices unlawfully segregate the Mexican-American students of the district.
As we noted above, this circuit has consistently taken the position that ability grouping of students is not, per se, unconstitutional. The merits of a program which places students in classrooms with others perceived to have similar abilities are hotly debated by educators; nevertheless, it is educators, rather than courts, who are in a better position ultimately to resolve the question whether such a practice is, on the whole, more beneficial than detrimental to the students involved. Thus, as a general /ule, school systems are free to employ ability grouping, even when such a policy has a segregative effect, so long, of course, as such a practice is genuinely motivated by educational concerns and not discriminatory motives. However, in school districts which have a past history of unlawful discrimination and are in the process of converting to a unitary school system, or have only recently completed such a conversion, ability grouping is subject to much closer judicial scrutiny. Under these circumstances we have prohibited districts from employing ability grouping as a device for assigning students to schools or classrooms, United States v. Gadsden County School District, supra; McNeal v. Tate County School District, supra. The rationale supporting judicial proscription of ability grouping under these circumstances is two-fold. First, ability grouping, when employed in such transitional circumstances may perpetuate the effects of past discrimination by re-segregating, on the basis of ability, students who were previously segregated in inferior schools on the basis of race or national origin. Second, a relatively recent history of discrimination may be probative evidence of a discriminatory motive which, when coupled with evidence of the segregative effect of ability grouping practices, may support a finding of unconstitutional discrimination.
Thus, in a case where the ability grouping practices of a school system are challenged, the court must always consider the history of the school system involved. If the system has no history of discrimination, or, if despite such a history, the system has achieved unitary status and maintained such status for a sufficient period of time that it seems reasonable to assume that any racially disparate impact of the ability grouping does not reflect either the lingering effects of past segregation or a contemporary segregative intent, then no impermissible racial classification is involved and ability grouping may be employed despite segregative effects. However, if the district’s history reveals a story of unremedied discrimination, or remedies of a very recent vintage which may not yet be fully effective to erase the effects of past discrimination, then the courts must scrutinize the effects of ability grouping with “punctilious care.” McNeal v. Tate County School Dis*997triet, id. at 1020. Even under these circumstances, however ability grouping is not always impermissible. If the statistical results of the ability grouping practices do not indicate “abnormal or unusual” segregation of students along racial lines, the practice is acceptable even in a system still pursuing desegregation efforts. Morales v. Shannon, supra at 414.
Despite the absence of district court findings on the questions whether RISD has a history of discrimination against Mexican-Americans and whether any past discrimination has been fully remedied, we are able to consider the merits of plaintiffs’ ability grouping claim insofar as it challenges the practices employed in grades 9-12. We note, first, that although different high school courses in Raymondville may be designed to accommodate students of different abilities or interests, self-selection, by students and parents, plays a very large part in the process by which students end up in a particular course. In light of this fact, we cannot conclude that “ability grouping,” insofar as that term refers to the practice of a school in assigning a student to a particular educational program designed for individuals of particular ability or achievement, is, in fact, employed at the high school level.
The district court’s failure to make findings concerning the RISD’s history does, however, severely handicap our review of the ability grouping practices employed in the central campus elementary school and the junior high school. RISD contends that we should deem these practices unobjectionable because even if the district court were to find that RISD has a history of unlawful discrimination, the effects of which have not yet been fully and finally remedied, the statistical results of RISD’s ability grouping practices, are, like the results of the ability grouping employed in Morales v. Shannon, supra, “not so abnormal or unusual ... as to justify an inference of discrimination.” Id. at 414. We cannot agree. In Morales, the overall student population in the grades where ability grouping was practiced was approximately 60% Mexican-American and 40% Anglo; however, approximately 61% of the students assigned to “high” groups were Anglo. Thus, 1.5 times as many Anglos were assigned to high groups as were enrolled in these grades as a whole. In Raymondville, the statistical results of the ability grouping are definitely more marked. For example, in grades kindergarten through three, during the academic year 1977-78, Anglo students formed approximately 17% of the student population at the central elementary campus; however 41% of the students in “high” ability classes for those grades were Anglo. Thus, there were approximately 2.4 times as many Anglos in high ability classes as there were in these grades as a whole. The figures in the upper grades for this year are comparable. In grades 4 and 5, there were approximately 2.3 times as many Anglos in high ability classes as in these grades as a whole; and in the junior high school grades 6-8, there were approximately 2.6 times as many Anglos in high groups as in the junior high school as a whole.
Statistical results such as these would not be permissible in a school system which has not yet attained, or only very recently attained, unitary status. Thus it is essential to examine the history of the RISD in order to determine the merits of the plaintiffs’ claims. On remand, therefore, the district court should reconsider the plaintiffs’ allegation that the ability grouping practices of the RISD are unlawful, insofar as grades K-8 are concerned, in light of the conclusions it reaches concerning the history of the district and the question whether it currently operates a unitary school system. If the district court finds that RISD has a past history of discrimination and has not yet maintained a unitary school system for a sufficient period of time that the effects of this history may reasonably be deemed to have been fully erased, the district’s current practices of ability grouping are barred because of their markedly segregative effect.
The historical inquiry is not, however, the only one that the district court must make on remand in order to determine the merits of the plaintiffs’ claims that RISD’s ability *998grouping practices are unlawful. The record suggests that in Raymondville “ability grouping” is intertwined with the district’s language remediation efforts and this intersection raises questions not present in our earlier cases involving ability grouping. The record indicates that the primary “ability” assessed by the district’s ability grouping practices in the early grades is the English language proficiency of the students. Students entering RISD kindergarten classes are given a test to determine whether their dominant language is English or Spanish. Predominantly Spanish speaking children are then placed in groups designated “low” and receive intensive bilingual instruction. “High” groups are those composed of students whose dominant language is English. “Ability groups” for first, second and third grade are determined by three basic factors: school grades, teacher recommendations and scores on standardized achievement tests. These tests are administered in English and cannot, of course, be expected to accurately assess the “ability” of a student who has limited English language skills and has been receiving a substantial part of his or her education in another language as part of a bilingual education program.
Nothing in our earlier cases involving ability grouping circumscribes the discretion of a school district, even one having a prior history of segregation, in choosing to group children on the basis of language for purposes of a language remediation or bilingual education program. Even though such a practice would predictably result in some segregation, the benefits which would accrue to Spanish speaking students by remedying the language barriers which impede their ability to realize their academic potential in an English language educational institution may outweigh the adverse effects of such segregation.4 See McNeal v. Tate County School District, supra at 1020 (ability grouping may be permitted in a school district with a history of segregation “if the district can demonstrate that its assignment method is not based on the present results of past segregation or will remedy such results through better educational opportunities.”)
Language grouping is, therefore, an unobjectionable practice, even in a district with a past history of discrimination. However, a practice which actually groups children on the basis of their language ability and then identifies these groups not by a description of their language ability but with a general ability label is, we think, highly suspect. In a district with a past history of discrimination, such a practice clearly has the effect of perpetuating the stigma of inferiority originally imposed on Spanish speaking children by past practices of discrimination. Even in the absence of such a history, we think that if the district court finds that the RISD’s ability grouping practices operate to confuse measures of two different characteristics, /. e., language and intelligence, with the result that predominantly Spanish speaking children are inaccurately labeled as “low ability,” the court should consider the extent to which such an irrational procedure may in and of itself be evidence of a discriminatory intent to stigmatize these children as inferior on the basis of their ethnic background.
III. TEACHERS
Testimony given in both the administrative proceeding and the trial of this civil suit indicates that the relatively small number of Mexican-American teachers and administrators employed by the Raymondville school district is a matter of great concern to Mexican-American students and their parents. Many persons in the community apparently believe that the disparity between the percentage of teachers in the district who are Mexican-American, 27%, and the percentage of students who are *999Mexican-American, 88%, is one of the major reasons for the underachievement and high dropout rate of Mexican-American students in Raymondville. Plaintiffs urge that this statistical disparity is both the result of, and evidence of, unlawful discrimination by RISD. The school district insists that it shares this desire to see more Mexican-American teachers employed in Raymond-ville schools, and argues that the current situation is not the result of unlawful discrimination on its part, but rather a reflection of the fact that certain characteristics of Raymondville, notably the lack of cultural activities and housing, make it difficult to recruit Mexican-American teachers, who are actively sought by many other school districts in Texas. The district court agreed with the RISD’s contentions and concluded that the school district did not discriminate against Mexican-Amerieans in either the hiring or promotion of teachers or administrators. In order to review the merits of that conclusion, we think it appropriate to examine first the precise legal basis for the teacher discrimination claim advanced by the plaintiffs in order to discern the correct legal framework for our review.
At the outset we note that the question whether RISD discriminates in the employment or promotion of teachers or administrators reaches us in a somewhat unusual posture. The class of plaintiffs in this case includes only Mexican-American students and their parents; no RISD employee, former employee or applicant for employment by the district is a party to this suit. Although students and parents are not typically the persons who bring suit to remedy alleged discrimination in the hiring and promotion of teachers and administrators in a school district, we do not believe they lack standing to do so. Plaintiffs premise their claim on the fourteenth amendment, and 42 U.S.C. § 1983, Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d and the Equal Educational Opportunity Act, 20 U.S.C. § 1701 et seq. The Equal Educational Opportunities Act (EEOA) explicitly provides in § 1703(d) that “discrimination by an educational agency on the basis of race, color or national origin in the employment ... of faculty or staff” constitutes a denial of equal educational opportunity. The statute also expressly provides a private right of action for persons denied such an “equal educational opportunity” in § 1706. Thus the class of students here clearly have standing to complain of, and a private cause of action for relief from, alleged discrimination by RISD in the hiring and promotion of teachers and staff under this statute.
With regard to the plaintiffs’ rights to assert a claim based upon this type of discrimination under the constitution and Title VI, we note that historically, dual school systems were maintained not only by segregation of students on the basis of race but also through discrimination in hiring and assignment of teachers. Consequently, as part of the remedy ordered in school desegregation cases, we have often included a provision intended to assure that a school district did not perpetuate unlawful school segregation through discriminatory employment practices.5 Such remedial orders implicitly acknowledge that the Equal Protection Clause, which outlaws discrimination on the basis of race or national origin in public education, requires not only that students shall not themselves be discriminated against on the basis of race by assignment to a particular school or classroom, but that they shall not be deprived of an equal educational opportunity by being forced to receive instruction from a faculty and administration composed of persons selected on the basis of unlawful racial or ethnic crite*1000ria. Thus, we think that the class of plaintiffs here may also assert a cause of action based upon unconstitutional racial discrimination in employment of teachers and administrators under 42 U.S.C. § 1983. In making this claim, the students are not attempting to vindicate the constitutional rights of the teachers involved but only seeking to remedy a denial of equal protection they claim to have suffered as a result of faculty discrimination. They have thus suffered an “injury in fact” and have shown a “sufficient personal stake in the outcome of the controversy” to establish their standing to assert a claim that RISD discriminates in its employment practices. Tasby v. Estes, 634 F.2d 1103 (5th Cir. 1981); Otero v. Mesa Valley School District No. 51, 568 F.2d 1312, 1314 (10th Cir. 1977) (quoting Sierra Club v. Morton, 405 U.S. 727, 92 S.Ct. 1361, 31 L.Ed.2d 636 (1976)).
With regard to Title VI, although the Supreme Court has never explicitly so held, there is authority in this circuit acknowledging a private right of action under this statute. Bossier Parish School Board v. Lemon, 370 F.2d 847, 852-51 (5th Cir.), cert. denied, 388 U.S. 911, 87 S.Ct. 2116, 18 L.Ed.2d 1350 (1967). In any event, since a majority of the Court has now taken the position that Title VI proscribes the same scope of classifications based on race as does the Equal Protection Clause, University of California Regents v. Bakke, 438 U.S. 265, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978), the question whether plaintiffs have an independent cause of action under that statute is not a significant one in this case.
Having concluded that the plaintiffs in this case have standing and a cause of action to complain of discrimination by RISD in the employment of faculty and staff, we turn to examine more carefully the elements of this cause of action and the proof adduced by the plaintiffs in support of their claim. With regard to the plaintiffs’ claims based upon Title VI and the Equal Protection Clause, we note that it is now well-established that in order to assert a claim based upon unconstitutional racial discrimination a party must not only allege and prove that the challenged conduct had a differential or disparate impact upon persons of different races, but also assert and prove that the governmental actor, in adopting or employing the challenged practices or undertaking the challenged action, intended to treat similarly situated persons differently on the basis of race. Personnel Administrator of Massachusetts v. Feeney, 442 U.S. 256, 99 S.Ct. 2282, 60 L.Ed.2d 870 (1979); Village of Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977); Washington v. Davis, 426 U.S. 229, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976). Thus, discriminatory intent, as well as disparate impact, must be shown in employment discrimination suits brought against public employers under Title VI, 42 U.S.C. § 1981 or § 1983. Lee v. Conecuh County Board of Education, 634 F.2d 959 (5th Cir. 1981); Lee v. Washington County Board of Education, 625 F.2d 1235 (5th Cir. 1980); Crawford v. Western Electric Co., Inc., 614 F.2d 1300 (5th Cir. 1980); Williams v. DeKalb County, 582 F.2d 2 (5th Cir. 1978). By contrast, in an employment discrimination action premised upon Title VII, a party may rely solely upon the disparate impact theory of discrimination recognized in Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971). To establish a cause of action based upon this theory, no intent to discriminate need be shown.
The question of what constitutes “discrimination” in the employment practices of a school district within the meaning of § 1703(d) of the EEOA, specifically the question whether intent is required in order to establish a cause of action for discrimination under that statute, cannot be so easily answered by reference to established judicial interpretations of the statute. There is little judicial precedent construing this provision. After examining carefully the language and legislative history of the statute, we have, however, reached the conclusion that the discriminatory conduct proscribed by § 1703(d) is coextensive with that prohibited by the fourteenth amendment and Title VI and does not encompass conduct *1001which might violate Title VII because, although not motivated by racial factors, it has a disparate impact upon persons of different races. Certain of the subsections of § 1703 which define the practices which constitute a denial of equal educational opportunity, explicitly include only intentional or deliberate acts. For example, § 1703(a) prohibits “deliberate segregation ... on the basis of race, color or national origin ...” and § 1703(e) bans transfers of students which have “the purpose and effect” of increasing segregation. The language of 1703(d) refers only to “discrimination” and does not contain such an explicit intent requirement. In considering the EEOA under different circumstances, we have found that some of its provisions “go beyond the acts and practices proscribed prior to the EEOA’s passage” and that by its terms, the statute explicitly makes unlawful practices, such as segregation of students on the basis of sex, which may not violate the fourteenth amendment because of the lesser scrutiny given six-based classifications under the Equal Protection Clause, United States v. Hinds County School Board, 560 F.2d 619 (5th Cir. 1977). Although by language in the act explicitly prohibiting segregation on the basis of sex in pupil assignments Congress clearly evidenced an intent that the statute prohibit certain types of conduct not unlawful under the Constitution, we have found no evidence to suggest that the particular subsection which concerns us here, § 1703(d), was designed to encompass a broader variety of employment practices than the provisions of the fourteenth amendment or Title VI. As other courts confronted with the task of interpreting the EEOA have noted, the legislative history of this statute is very sparse, indeed almost non-existent. Guadalupe Organization, Inc. v. Tempe Elementary School Dist. No. 3, 587 F.2d 1022 (9th Cir. 1978). The EEOA was a floor amendment to the 1974 legislation amending the Elementary and Secondary Education Act of 1965, 88 Stat. 338-41, 346-48, 352 (codified in scattered sections of 20 U.S.C.). We agree with the Guadalupe court’s suggestion that “[t]he interpretation of floor amendments unaccompanied by illuminating debate should adhere closely to the ordinary meaning of the amendment’s language.” 587 F.2d at 1030. Unlike Title VII there is nothing in the language of § 1703(d) to suggest that practices having only disparate impact, as well as those motivated by a discriminatory animus, were to be prohibited. Title VII, unlike § 1703(d), makes it an unlawful practice for an employer not only to “discriminate” against individuals on the basis of certain criteria but also makes it unlawful “to limit, segregate or classify [persons] in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee because of ... race, color, religion, sex or national origin.” It is this latter provision, which was interpreted in Griggs to prohibit facially neutral practices having a disparate impact on persons of different races. No similar provision or description of employment practices having a disparate impact was included in the Equal Educational Opportunities Act. Thus, we conclude that the elements of plaintiff’s cause of action for discrimination in the hiring and promotion of teachers and administrators under the Equal Educational Opportunities Act are the same as the elements of their claims premised on the fourteenth amendment and § 1983 and Title VI.
Although the question whether RISD unlawfully discriminates against Mexican-Americans in the hiring or promotion of faculty and administrators reaches us in the somewhat unusual posture of a case brought by students, we think the legal analysis of their claim is properly drawn from the approach used to assess the merits of more traditional class action and pattern and practice employment discrimination suits. In civil rights cases generally we have noted that a district court’s finding of discrimination or no discrimination is a determination of an ultimate fact; thus, we must make an independent determination of this question. Phillips v. Joint Legislative Committee, 637 F.2d 1014, 1024-25 (5th Cir. 1981); Danner v. U.S. Civil Service *1002Commission, 635 F.2d 427 (5th Cir. 1981); Thompson v. Leland Police Dep’t, 633 F.2d lili (5th Cir. 1980); Shepard v. Beaird-Poulan, Inc., 617 F.2d 87 (5th Cir. 1980); Ramirez v. Sloss, 615 F.2d 163 (5th Cir. 1980). In undertaking such an independent review, however, we are bound by the subsidiary factual determinations that the district court made in the course of considering the ultimate issue of discrimination, unless these subsidiary findings are clearly erroneous within the meaning of Fed.R. Civ.P. 52(a). In this case, the district court apparently based its conclusion that RISD did not discriminate against Mexican-Americans in the hiring or promotion of teachers or administrators on subsidiary findings that: (1) RISD currently hires a higher percentage of Mexican-American applicants for teaching positions than Anglo applicants; (2) the school district hires many teachers from nearby universities which have substantial numbers of Mexican-American students; and (3) the school district has a difficult time recruiting Mexican-American teachers because, although its salaries are commensurate with those paid by other schools in the area, Raymond-ville has very limited housing and cultural activities. Although we do not characterize any of these subsidiary findings as clearly erroneous, we do not believe they are sufficient to support an ultimate finding that RISD does not discriminate against Mexican-Americans in the employment of teachers or administrators.
In class action or pattern and practice employment discrimination suits, the question whether the employer discriminates against a particular group in making hiring decisions requires, as a first and fundamental step, a statistical comparison between the racial composition of the employer’s work force and that of the relevant labor market. In many of these cases the nature of the jobs involved suggests that the relevant labor market is coextensive with the general population in the geographical areas from which the employer might reasonably be expected to draw his work force. Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977); Markey v. Tenneco Oil Co., 635 F.2d 497 (5th Cir. 1981); United States v. City of Alexandria, 614 F.2d 1358, 1364 (5th Cir. 1980). In this case, plaintiffs have relied heavily on the disparity between the percentage of the Raymondville school population consisting of Mexican-Americans (approximately 85%) and the percentage of the faculty in the Raymondville schools who are Mexican-American (27%), in support of their contention that RISD discriminates in its employment decisions. Plaintiffs urge that this statistical disparity coupled with the evidence of a past history of segregation in the Raymondville schools sufficed to make out a prima facie case of discrimination which shifted to the defendants a heavy burden of rebuttal which they failed to meet.
We think the plaintiffs’ suggested comparison is not the relevant one. Where, as here, the nature of the employment involved suggests that the pool of people qualified to fill the positions is not likely to be substantially congruent with the general population, the relevant labor market must be separately and distinctly defined. In Hazelwood School District v. United States, 433 U.S. 299, 97 S.Ct. 2736, 53 L.Ed.2d 768 (1977), the Supreme Court considered the question of how to define the relevant labor pool in a case involving a claim that a school district engaged in a pattern and practice of employment discrimination in the hiring of teachers. The Court disapproved of the comparison, which had been made by the district court, between the racial composition of the district’s teacher work force and the student population. Such an approach, the Court admonished, “fundamentally misconceived the role of statistics in employment discrimination cases.” Id. at 308,97 S.Ct. at 2741-42. The proper comparison in a case involving school teachers was
between the racial composition of [the district’s] teaching staff and the racial composition of the qualified public school teacher population in the relevant labor market.
Id.
The district court’s memorandum opinion in this case does not indicate that any such *1003comparison was made here. The district court did apparently compare the data concerning the ethnic composition of the pool of persons who applied for teaching positions at Raymondville, with the ethnic composition of the persons hired. The court found that a larger percentage of Mexican-American applicants than Anglos was hired. The record also indicates that Mexican-Americans comprise a larger percentage of the teachers hired in RISD than they do of the applicant pool. In the usual hiring discrimination case this type of applicant flow data provides a very good picture of the relevant labor market because it allows one to compare the ethnic composition of an employer’s workforce with that of the pool of persons actually available for hire by the employer. Markey, supra, at 499. However, in cases such as this one where there is an allegation that the employer’s discriminatory practices infect recruiting, the process by which applications are solicited, such applicant flow data cannot be taken at face value and assumed to constitute an accurate picture of the relevant labor market. Discriminatory recruiting practices may skew the ethnic composition of the applicant pool. B. L. Schlei and P. Grossman, Employment Discrimination Law, 445 (1976).
In a case such as this one, the relevant labor market must first be defined separately from the applicant pool in order to determine the merits of the claim of discrimination in recruiting. A statistically significant disparity between the racial composition of the applicant pool and that of the relevant labor market may create a prima facie case of discrimination in recruiting. Because determination of the relevant labor market, the geographical area from which we might reasonably expect RISD to draw applicants and teachers, and of the ethnic composition of the group of persons qualified for teaching positions in this area, is an essentially factual matter within the special competence of the district court, Hazelwood, supra at 312, 97 S.Ct. at 2744, Markey, supra at 498, we remand the issue of discrimination in teacher hiring to the district court for further findings in accordance with the analysis the Supreme Court delineated in Hazelwood and which we have employed in class action and pattern and practice employment discrimination suits. See, e. g., Phillips v. Joint Legislative Committee, supra at 1024-25; Markey, supra; E.E.O.C. v. Datapoint Corporation, 570 F.2d 1264 (5th Cir. 1978).
With regard to the question whether RISD discriminates in the hiring or promotion of persons to administrative positions in the district, the district court concluded that there was no discrimination in this area. In recent years, the percentage of Mexican-Americans serving in administrative positions in the Raymondville School District has been roughly comparable to the percentage of Mexican-Americans on the faculty. For example in 1976, Mexican-Americans occupied 5 of the 16 administrative positions in the district (24%); in the same year 26% of the district’s teachers were Mexican-American. Given the small numbers involved we are not prepared to term this a significant disparity. The record indicates that, as a general rule, the RISD prefers to hire administrative personnel from within the ranks of its current employees; thus the statistical evidence in this case would not seem to support an inference of discrimination in promotion, unless, of course, discrimination in hiring is established. In that case, the district court should, on remand, reconsider the issue of discrimination in promotion as well.
The comparison of the employment statistics of RISD with the ethnic composition of the relevant labor market goes to the determination whether the plaintiff made out a prima facie case of unlawful discrimination. If, on remand, the district court concludes that plaintiffs succeeded in making out a prima facie case, the court should determine the nature and weight of the burden of rebuttal this prima facie case placed on the RISD. As we noted above, that burden may differ depending on the conclusions the district court reaches concerning the district’s history. See text supra, at 994-996.
*1004The district court must, of course, then consider whether RISD adduced evidence sufficient to rebut the plaintiffs’ prima facie case, i. e., evidence tending to suggest that the statistical underrepresentation of Mexican-Americans established by the plaintiffs’ prima facie case was not the result of intentional discrimination by the school district. We note that RISD has urged that since Mexican-Americans from a majority of the voting population in the school district, are present on the district’s board and have, along with the Anglo majority of the board, voted for and approved most of the hiring and promotion decisions which the plaintiffs have challenged here, the district has adequately rebutted any inference of discriminatory intent which might be raised by plaintiffs’ prima facie case.
Although there have been Mexican-American members on the RISD board, there is no evidence in the record that Mexican-Americans have ever formed a majority of the board. Further, the school board’s role in the teacher employment process appears to be a largely ministerial one. From the minutes of the school board meetings contained in the record, it appears that the school board does not itself receive and review the files of all applicants or involve itself in the recruiting process. The minutes suggest that the superintendent presents a slate of teachers to the board for its formal approval en masse. Thus, the record suggests that the school board has delegated primary responsibility for the recruitment and hiring of teachers and administrators to the superintendent, a position which has always been occupied by an Anglo. This suggests the possibility that the Mexican-Americans on the board may not, in fact, be in a position to exercise much power over the district’s employment decisions.
In any event, the Supreme Court has rejected the argument that this type of “governing majority” theory can, standing alone, rebut a prima facie case of intentional discrimination. In Castaneda v. Partida, 430 U.S. 482, 97 S.Ct. 1272, 51 L.Ed.2d 498 (1977), the Supreme Court considered a similar argument. Castaneda involved a challenge by a Mexican-American to the grand jury selection procedures employed in Hidalgo County, Texas. The state argued that the plaintiffs’ prima facie case of intentional discrimination, which consisted of statistical evidence of a significant under-representation of Mexican-Americans on grand juries, was effectively rebutted merely by evidence that Mexican-Americans were an effective political majority in the county and occupied many county offices, including three of the five grand jury commissioners’ posts. The state reasoned that these facts made it highly unlikely that Mexican-Americans were being intention ally excluded from the county’s grand juries. The Supreme Court, however, held that such a governing majority theory could not, standing alone, discharge the burden placed on the defendants by plaintiffs’ prima facie case. This is not, of course, to say that such evidence is not relevant as part of the district’s rebuttal, but only that it may not be deemed conclusive.
We express no opinion as to the outcome of the inquiry which we have directed the district court to make. The question of whether the plaintiffs have made out a prima facie case of unlawful discrimination in the employment practices of the district and the question of whether that case, if made out, has been adequately rebutted are reserved to the district court in the first instance.
IV. THE BILINGUAL EDUCATION AND LANGUAGE REMEDIATION PROGRAMS OF THE RAYMOND-VILLE SCHOOLS6
RISD currently operates a bilingual education program for all students in kinder*1005garten through third grade.7 The language ability of each student entering the Raymondville program is assessed when he or she enters school. The language dominance test currently employed by the district is approved for this purpose by the TEA. The program of bilingual instruction offered students in the Raymondville schools has been developed with the assistance of expert consultants retained by the TEA and employs a group of materials developed by a regional educational center operated by the TEA. The articulated goal of the program is to teach students fundamental reading and writing skills in both Spanish and English by the end of third grade.
Although the program’s emphasis is on the development of language skills in the two languages, other cognitive and substantive areas are addressed, e. g., mathematics skills are taught and tested in Spanish as well as English during these years. All of the teachers employed in the bilingual education program of the district have met the minimum state requirements to teach bilingual classes. However, only about half of these teachers are Mexican-American and native Spanish speakers; the other teachers in the program have been certified to teach bilingual classes following a 100 hour course designed by TEA to give them a limited Spanish vocabulary (700 words) and an understanding of the theory and methods employed in bilingual programs. Teachers in the bilingual program are assisted by classroom aides, most of whom are fluent in Spanish.
RISD does not offer a formal program of bilingual education after the third grade. In grades 4 and 5, although classroom instruction is only in English, Spanish speaking teacher aides are used to assist students having language difficulties which may impair their ability to participate in classroom activities. For students in grades 4-12 having limited English proficiency or academic deficiencies in other areas, the RISD provides assistance in the form of a learning center operated at each school. This center provides a diagnostic/prescriptive program in which students’ particular academic deficiencies, whether in language or other areas, are identified and addressed by special remedial programs. Approximately 1,000 of the district’s students, almost one-third of the total enrollment, receive special assistance through small classes provided by these learning centers. The district also makes English as a Second Language classes and special tutoring in English available to all students in all grades; this program is especially designed to meet the needs of limited English speaking students who move into the district in grades above 3.8
*1006Plaintiffs claim that the bilingual education and language remediation programs offered by the Raymondville schools are educationally deficient and unsound and that RISD’s failure to alter and improve these programs places the district in violation of Title VI and the Equal Educational Opportunities Act. The plaintiffs claim that the RISD programs fail to comport with the requirements of the “Lau Guidelines” promulgated in 1975 by the Department of Health, Education and Welfare. Specifically, plaintiffs contend that the articulated goal of the Raymondville program — to teach limited English speaking children to read and write in both English and Spanish at grade level— is improper because it overemphasizes the development of English language skills to the detriment of the child’s overall cognitive development. Under the Lau Guidelines, plaintiffs argue, “pressing English on the child is not the first goal of language remediation.” Plaintiffs criticize not only the premise and purpose of the RISD language programs but also particular aspects of the implementation of the program. Specifically, plaintiffs take issue with the tests the district employs to identify and assess limited English speaking children and the qualifications of the teachers and staff involved in the district’s language remediation program. Plaintiffs contend that in both of these areas RISD falls short of standards established by the Lau Guidelines and thus has fallen out of compliance with Title VI and the EEOA.
We agree with the district court that RISD’s program does not violate Title VI. Much of the plaintiffs’ argument with regard to Title VI is based upon the premise that the Lau Guidelines are administrative regulations applicable to the RISD and thus should be given great weight by us in assessing the legal sufficiency of the district’s programs. This premise is, however, flawed. The Department of HEW, in assessing the district’s compliance with Title VI, acknowledged that the Lau Guidelines were inapplicable to an evaluation of the legal sufficiency of the district’s language program. The Lau Guidelines were formulated by the Department following the Supreme Court’s decision in Lau v. Nichols, 414 U.S. 563, 94 S.Ct. 786, 39 L.Ed.2d 1 (1974). In Lau, the Supreme Court determined that a school district’s failure to provide any English language assistance to substantial numbers of non-English speaking Chinese students enrolled in the district’s schools violated Title VI because this failure denied these students “a meaningful opportunity to participate in the educational program” offered by the school district, 414 U.S. at 568, 94 S.Ct. at 789. Lau involved a school district which offered many non-English speaking students no assistance in developing English language skills; in declaring such an omission unlawful, the Court did not dictate the form such assistance must take. Indeed the Court specifically noted that the school district might undertake any one of several permissible courses of language remediation:
Teaching English to the students of Chinese ancestry who do not speak the language is one choice. Giving instruction to this group in Chinese is another. There may be others.
Id. at 565, 94 S.Ct. at 787. The petitioners in Lau did not specifically request, nor did the Court require, court ordered relief in the form of bilingual education; the plaintiffs in that case sought only “that the Board of Education be directed to apply its expertise to the problem .... ” Id.
Following the Supreme Court’s decision in Lau, HEW developed the “Lau Guidelines” as a suggested compliance plan for school districts which, as a result of Lau, were in violation of Title VI because they failed to provide any English language assistance to students having limited English proficiency. Clearly, Raymondville is not culpable of such a failure. Under these *1007circumstances, the fact that Raymondville provides (and long has provided) a program of language remediation which differs in some respects from these guidelines is, as the opinion of the Reviewing Authority for the OCR noted, “not in itself sufficient to rule that program unlawful in the first instance.”
The Lau Guidelines were the result of a policy conference organized by HEW; these guidelines were not developed through the usual administrative procedures employed to draft administrative rules or regulations. The Lau Guidelines were never published in the Federal Register. Since the Department itself in its administrative decision found that RISD’s departure from the Lau Guidelines was not determinative of the question whether the district complied with Title VI, we do not think that these guidelines are the sort of administrative document to which we customarily give great deference in our determinations of compliance with a statute.
We must confess to serious doubts not only about the relevance of the Lau Guidelines to this case but also about the continuing vitality of the rationale of the Supreme Court’s opinion in Lau v. Nichols which gave rise to those guidelines. Lau was written prior to Washington v. Davis, 426 U.S. 229, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976), in which the Court held that a discriminatory purpose, and not simply a disparate impact, must be shown to establish a violation of the Equal Protection Clause, and University of California Regents v. Bakke, 438 U.S. 265, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978), in which, as we have already noted, a majority of the court interpreted Title VI to be coextensive with the Equal Protection Clause. Justice Brennan’s opinion (in which Justices White, Marshall and Blackmun joined) in Bakke explicitly acknowledged that these developments raised serious questions about the vitality of Lau.
We recognize that Lau, especially when read in light of our subsequent decision in Washington v. Davis, 426 U.S. 229 [96 S.Ct. 2040, 48 L.Ed.2d 597] (1976), which rejected the general propostion that governmental action is unconstitutional solely because it has a racially disproportionate impact, may be read as being predicated upon the view that, at least under some circumstances, Title VI proscribes conduct which might not be prohibited by the Constitution. Since we are now of the opinion, for the reasons set forth above, that Title Vi’s standard, applicable alike to public and private recipients of federal funds, is no broader than the Constitution’s, we have serious doubts concerning the correctness of what appears to be the premise of that decision.
Id. at 352, 98 S.Ct. at 2779. Although the Supreme Court in Bakke did not expressly overrule Lau, as we noted above, we understand the clear import of Bakke to be that Title VI, like the Equal Protection Clause, is violated only by conduct animated by an intent to discriminate and not by conduct which, although benignly motivated, has a differential impact on persons of different races. Whatever the deficiencies of the RISD’s program of language remediation may be, we do not think it can seriously be asserted that this program was intended or designed to discriminate against Mexican-American students in the district. Thus, we think it cannot be said that the arguable inadequacies of the program render it violative of Title VI.
Plaintiffs, however, do not base their legal challenge to the district’s language program solely on Title VI. They also claim that the district’s current program is unlawful under § 1703(f) of the EEOA which makes it unlawful for an educational agency to fail to take “appropriate action to overcome language barriers that impede equal participation by its students in its instructional programs.” As we noted above in dissecting the meaning of § 1703(d) of the EEOA, we have very little legislative history from which to glean the Congressional intent behind the EEOA’s provisions. Thus, as we did in examining § 1703(d), we shall adhere closely to the plain language of § 1703(f) in defining the meaning of this provision. Unlike subsections (a) and (e) of § 1703, § 1703(f) does *1008not contain language that explicitly incorporates an intent requirement nor, like § 1703(d) which we construed above, does this subsection employ words such as “discrimination” whose legal definition has been understood to incorporate an intent requirement. Although we have not previously explicitly considered this question, in Morales v. Shannon, supra, we assumed that the failure of an educational agency to undertake appropriate efforts to remedy the language deficiencies of its students, regardless of whether such a failure is motivated by an intent to discriminate against those students, would violate § 1703(f) and we think that such a construction of that subsection is most consistent with the plain meaning of the language employed in § 1703(f). Thus, although serious doubts exist about the continuing vitality of Lau v. Nichols as a judicial interpretation of the requirements of Title VI or the fourteenth amendment, the essential holding of Lau, i. e., that schools are not free to ignore the need of limited English speaking children for language assistance to enable them to participate in the instructional program of the district, has now been legislated by Congress, acting pursuant to its power to enforce the fourteenth amendment, in § 1703(f).9 The difficult question presented by plaintiffs’ challenge to the current language remediation programs in RISD is really whether Congress in enacting § 1703(f) intended to go beyond the essential requirement of Lau, that the schools do something, and impose, through the use of the term “appropriate action” a more specific obligation on state and local educational authorities.
We do not believe that Congress, at the time it adopted the EEOA, intended to require local educational authorities to adopt any particular type of language remediation program. At the same time Congress enacted the EEOA, it passed the Bilingual Education Act of 1974, 20 U.S.C. § 880b et seq. (1976). The Bilingual Educational Act established a program of federal financial assistance intended to encourage local educational authorities to develop and implement bilingual education programs. The Bilingual Education Act implicitly embodied a recognition that bilingual education programs were still in experimental stages *1009and that a variety of programs and techniques would have to be tried before it could be determined which were most efficacious. Thus, although the Act empowered the U.S. Office of Education to develop model programs, Congress expressly directed that the state and local agencies receiving funds under the Act were not required to adopt one of these model programs but were free to develop their own. Conf.Rep. No. 93-1026, 93d Cong., 2nd Sess. (1974), reprinted in [1974] U.S.Code Cong. & Ad.News 4093, 4206.
We note that although Congress enacted both the Biligual Education Act and the EEOA as part of the 1974 amendments to the Elementary and Secondary Education Act, Congress, in describing the remedial obligation it sought to impose on the states in the EEOA, did not specify that a state must provide a program of “bilingual education” to all limited English speaking students. We think Congress’ use of the less specific term, “appropriate action,” rather than “biligual education,” indicates that Congress intended to leave state and local educational authorities a substantial amount of latitude in choosing the programs and techniques they would use to meet their obligations under the EEOA. However, by including an obligation to address the problem of language barriers in the EEOA and granting limited English speaking students a private right of action to enforce that obligation in § 1706, Congress also must have intended to insure that schools made a genuine and good faith effort, consistent with local circumstances and resources, to remedy the language deficiencies of their students and deliberately placed on federal courts the difficult responsibility of determining whether that obligation had been met.
Congress has provided us with almost no guidance, in the form of text or legislative history, to assist us in determining whether a school district’s language remediation efforts are “appropriate.” Thus we find ourselves confronted with a type of task which federal courts are ill-equipped to perform and which we are often criticized for undertaking — prescribing substantive standards and policies for institutions whose governance is properly reserved to other levels and branches of our government (i. e., state and local educational agencies) which are better able to assimilate and assess the knowledge of professionals in the field. Confronted, reluctantly, with this type of task in this case, we have attempted to devise a mode of analysis which will permit ourselves and the lower courts to fulfill the responsibility Congress has assigned to us without unduly substituting our educational values and theories for the educational and political decisions reserved to state or local school authorities or the expert knowledge of educators.
In a case such as this one in which the appropriateness of a particular school system’s language remediation program is challenged under § 1703(f), we believe that the responsibility of the federal court is threefold. First, the court must examine carefully the evidence the record contains concerning the soundness of the educational theory or principles upon which the challenged program is based. This, of course, is not to be done with any eye toward discerning the relative merits of sound but competing bodies of expert educational opinion, for choosing between sound but competing theories is properly left to the educators and public officials charged with responsibility for directing the educational policy of a school system. The state of the art in the area of language remediation may well be such that respected authorities legitimately differ as to the best type of educational program for limited English speaking students and we do not believe that Congress in enacting § 1703(f) intended to make the resolution of these differences the province of federal courts. The court’s responsibility, insofar as educational theory is concerned, is only to ascertain that a school system is pursing a program informed by an educational theory recognized as sound by some experts in the field or, at least, deemed a legitimate experimental strategy.
*1010The court’s second inquiry would be whether the programs and practices actually used by a school system are reasonably calculated to implement effectively the educational theory adopted by the school. We do not believe that it may fairly be said that a school system is taking appropriate action to remedy language barriers if, despite the adoption of a promising theory, the system fails to follow through with practices, resources and personnel necessary to transform the theory into reality.
Finally, a determination that a school system has adopted a sound program for alleviating the language barriers impeding the educational progress of some of its students and made bona fide efforts to make the program work does not necessarily end the court’s inquiry into the appropriateness of the system’s actions. If a school’s program, although premised on a legitimate educational theory and implemented through the use of adequate techniques, fails, after being employed for a period of time sufficient to give the plan a legitimate trial, to produce results indicating that the language barriers confronting students are actually being overcome, that program may, at that point, no longer constitute appropriate action as far as that school is concerned. We do not believe Congress intended that under § 1703(f) a school would be free to persist in a policy which, although it may have been “appropriate” when adopted, in the sense that there were sound expectations for success and bona fide efforts to make the program work, has, in practice, proved a failure.
With this framework to guide our analysis we now turn to review the district court’s determination that the RISD’s current language remediation programs were “appropriate action” within the meaning of § 1703(f). Implicit in this conclusion was a determination that the district had adequately implemented a sound program. In conducting this review, we shall consider this conclusion as a determination of a mixed question of fact and law. Therefore we shall be concerned with determining whether this conclusion was adequately supported by subsidiary findings of fact which do not appear clearly erroneous.
In this case, the plaintiffs’ challenge to the appropriateness of the RISD’s efforts to overcome the language barriers of its students does not rest on an argument over the soundness of the educational policy being pursued by the district, but rather on the alleged inadequacy of the program actually implemented by the district.10 Plaintiffs contend that in three areas essential to the adequacy of a bilingual program — curriculum, staff and testing Raymondville falls short. Plaintiffs contend that although RISD purports to offer a bilingual education program in grades K-3, the district’s curriculum actually overemphasizes the development of reading and writing skills in English to the detriment of education in other areas such as mathematics and science, and that, as a result, children whose first language was Spanish emerge from the bilingual education program behind their classmates in these other areas. The record in this case does not support plaintiffs’ allegation that the educational program for predominantly Spanish speaking students in grades K-3 provides significantly less attention to these other areas than does the curriculum used in the English language dominant classrooms. The bilingual education manual developed by the district outlines the basic classroom sched*1011ules for both Spanish dominant classrooms and English dominant classrooms. These schedules indicate that students in the Spanish language dominant classrooms spend almost exactly the same amount of classroom time on math, science and social studies as do their counterparts in the predominantly English speaking classrooms. The extra time that Spanish language dominant children spend on language development is drawn almost entirely from what might fairly be deemed the “extras” rather than the basic skills components of the elementary school curriculum, e. g., naps, music, creative writing and physical education.
Even if we accept this allegation as true, however, we do not think that a school system which provides limited English speaking students with a curriculum, during the early part of their school career, which has, as its primary objective, the development of literacy in English, has failed to fulfill its obligations under § 1703(f), even if the result of such a program is an interim sacrifice of learning in other areas during this period. The language of § 1703(f) speaks in terms of taking action “to overcome language barriers” which impede the “equal participation" of limited English speaking children in the regular instructional program. We believe the statute clearly contemplates that provision of a program placing primary emphasis on the development of English language skills would constitute “appropriate action.”
Limited English speaking students entering school face a task not encountered by students who are already proficient in English. Since the number of hours in any school day is limited, some of the time which limited English speaking children will spend learning English may be devoted to other subjects by students who entered school already proficient in English. In order to be able ultimately to participate equally with the students who entered school with an English language background, the limited English speaking students will have to acquire both English language proficiency comparable to that of the average native speakers and to recoup any deficits which they may incur in other areas of the curriculum as a result of this extra expenditure of time on English language development. We understand § 1703(f) to impose on educational agencies not only an obligation to overcome the direct obstacle to learning which the language barrier itself poses, but also a duty to provide limited English speaking ability students with assistance in other areas of the curriculum where their equal participation may be impaired because of deficits incurred during participation in an agency’s language remediation program. If no remedial action is taken to overcome the academic deficits that limited English speaking students may incur during a period of intensive language training, then the language barrier, although itself remedied, might, nevertheless, pose a lingering and indirect impediment to these students’ equal participation in the regular instructional program. We also believe, however, that § 1703(f) leaves schools free to determine whether they wish to discharge these obligations simultaneously, by implementing a program designed to keep limited English speaking students at grade level in other areas of the curriculum by providing instruction in their native language at the same time that an English language development effort is pursued, or to address these problems in sequence, by focusing first on the development of English language skills and then later providing students with compensatory and supplemental education to remedy deficiencies in other areas which they may develop during this period. In short, § 1703(f) leaves schools free to determine the sequence and manner in which limited English speaking students tackle this dual challenge so long as the schools design programs which are reasonably calculated to enable these students to attain parity of participation in the standard instructional program within a reasonable length of time after they enter the school system. Therefore, we disagree with plaintiffs’ assertion that a school system which chooses to focus first on English language development and later provides students with an intensive remedial program *1012to help them catch up in other areas of the curriculum has failed to fulfill its statutory obligation under § 1703(f).
Although we therefore find no merit in the plaintiffs’ claim that RISD’s language remediation programs are inappropriate under § 1703 because of the emphasis the curriculum allegedly places on English language development in the primary grades, we are more troubled by the plaintiffs’ allegations that the district’s implementation of the program has been severely deficient in the area of preparing its teachers for bilingual education. Although the plaintiffs raised this issue below and introduced evidence addressed to it, the district court made no findings on the adequacy of the teacher training program employed by RISD.11 We begin by noting that any school district that chooses to fulfill its obligations under § 1703 by means of a bilingual education program has undertaken a responsibility to provide teachers who are able competently to teach in such a program. The record in this case indicates that some of the teachers employed in the RISD bilingual program have a very limited command of Spanish, despite completion of the TEA course. Plaintiffs’ expert witness, Dr. Jose Cardenas, was one of the bilingual educators who participated in the original design of the 100 hour continuing education course given to teachers already employed in RISD in order to prepare them to teach bilingual classes. He testified that a subsequent evaluation of the program showed that although it was effective in introducing teachers to the methodology of bilingual education and preparing them to teach the cultural history and awareness components of the bilingual education program, the course, was “a dismal failure in the development of sufficient proficiency in a language other than English to qualify the people for teaching bilingual programs.” Although the witnesses familiar with the bilingual teachers in the Raymondville schools did not testify quite as vividly to the program’s inadequacy, testimony of those involved in the RISD’s program suggested that despite completion of the 100 hour course, some of the district’s English speaking teachers were inadequately prepared to teach in a bilingual classroom. Mr. Inez Ibarra, who was employed by the district as bilingual supervisor prior to his appointment to the principalship of L. C. Smith School in 1977, testified in the administrative hearing that he had observed the teachers in the bilingual program at Raymondville and that some of the teachers had difficulty communicating in Spanish in the classroom and that there were teachers in the program who taught almost exclusively in English, using Spanish, at most, one day per week. He also described the evaluation program used to determine the Spanish proficiency of the teachers at the end of the 100 hour course. Teachers were required to write a paragraph in Spanish. Since in completing this task, they were permitted to use a Spanish-English dictionary, Ibarra acknowledged that this was not a valid measure of their Spanish vocabulary. Teachers also read orally from a Spanish language text and answered oral questions addressed to them by the RISD certification committee. There was no formal grading of the examination; the certification committee had no guide to measure the Spanish language vocabulary of the teachers based on their performance on the exam. Thus, it may well have been impossible for the committee to determine whether the teachers had mastered even the 700 word vocabulary the TEA had deemed the minimum to enable a teacher to work effectively in a bilingual elementary classroom. Following the examination, the committee would have an informal discussion among themselves and decide whether or not the teacher was qualified. Mr. Ibarra testified that the certifi*1013cation committee had approved some teachers who were, in his opinion, in need of more training — “much more than what they were given.”
The record in this case thus raises serious doubts about the actual language competency of the teachers employed in bilingual classrooms by RISD and about the degree to which the district is making a genuine effort to assess and improve the qualifications of its bilingual teachers. As in any educational program, qualified teachers are a critical component of the success of a language remediation program. A bilingual education program, however sound in theory, is clearly unlikely to have a significant impact on the language barriers confronting limited English speaking school children, if the teachers charged with day-to-day responsibility for educating these children are termed “qualified” despite the fact that they operate in the classroom under their own unremedied language disability. The use of Spanish speaking aides may be an appropriate interim measure, but such aides cannot, RISD acknowledges, take the place of qualified bilingual teachers. The record in this case strongly suggests that the efforts RISD has made to overcome the language barriers confronting many of the teachers assigned to the bilingual education program are inadequate. On this record, we think a finding to the contrary would be clearly erroneous. Nor can there be any question that deficiencies in the in-service training of teachers for bilingual classrooms seriously undermine the promise of the district’s bilingual education program. Until deficiencies in this aspect of the program’s implementation are remedied, we do not think RISD can be deemed to be taking “appropriate action” to overcome the language disabilities of its students. Although we certainly hope and expect that RISD will attempt to hire teachers who are already qualified to teach in a bilingual classroom as positions become available, we are by no means suggesting that teachers already employed by the district should be replaced or that the district is limited to hiring only teachers who are already qualified to teach in a bilingual program. We are requiring only that RISD undertake further measures to improve the ability of any teacher, whether now or hereafter employed, to teach effectively in a bilingual classroom.
On the current record, it is impossible for us to determine the extent to which the language deficiencies of some members of RISD’s staff are the result of the inadequacies inherent in TEA’s 100 hour program (including the 700 word requirement which may be an insufficient vocabulary) or the extent to which these deficiencies reflect a failure to master the material in that course. Therefore, on remand, the district court should attempt to identify more precisely the cause or causes of the Spanish language deficiencies experienced by some of the RISD’s teachers and should require both TEA and RISD to devise an improved in-service training program and an adequate testing or evaluation procedure to assess the qualifications of teachers completing this program.12
The third specific area in which plaintiffs claim that RISD programs are seriously deficient is in the testing and evaluation of students having limited English proficiency. Plaintiffs claim first that the language dominance placement test used to evaluate students entering Raymondville schools is inadequate. Although it appears that at the time of the administrative hearing in this case, RISD was not employing one of the language tests approved by the TEA, by the time of the trial in this civil suit RISD had adopted a test approved for this purpose by TEA. None of plaintiffs’ expert witnesses testified that this test was an inappropriate one.13 Thus, we do not think *1014there is any reason to believe that the district is deficient in the area of initial evaluation of students entering the bilingual program.
A more difficult question is whether the testing RISD employs to measure the progress of students in the bilingual education program is adequate. Plaintiffs, contend, RISD apparently does not deny, and we agree that proper testing and evaluation is essential in determining the progress of students involved in a bilingual program and ultimately, in evaluating the program itself. In their brief, plaintiffs contend that RISD’s testing program is inadequate because the limited English speaking students in the bilingual program are not tested in their own language to determine their progress in areas of the curriculum other than English language literacy skills. Although during the bilingual program Spanish speaking students receive much of their instruction in these other areas in the Spanish language, the achievement level of these students is tested, in part, by the use of standardized English language achievement tests. No standardized Spanish language tests are used. Plaintiffs contend that testing the achievement levels of children, who are admittedly not yet literate in English and are receiving instruction in another language, through the use of an English language achievement test, does not meaningfully assess their achievement, any more than it does their ability, a contention with which we can scarcely disagree.
Valid testing of students’ progress in these areas is, we believe, essential to measure the adequacy of a language remediation program. The progress of limited English speaking students in these other areas of the curriculum must be measured by means of a standardized test in their own language because no other device is adequate to determine their progress vis-a-vis that of their English speaking counterparts. Although, as we acknowledged above, we do not believe these students must necessarily be continuously maintained at grade level in other areas of instruction during the period in which they are mastering English, these students cannot be permitted to incur irreparable academic deficits during this period. Only by measuring the actual progress of students in these areas during the language remediation program can it be determined that such irremediable deficiencies are not being incurred. The district court on remand should require both TEA and RISD to implement an adequate achievement test program for RISD in accordance with this opinion. If, following the district court’s inquiry into the ability grouping practices of the district, such practices are allowed to continue, we assume that Spanish language ability tests would be employed to place students who have not yet mastered the English language satisfactorily in ability groups.
Finally plaintiffs contend that test results indicate that the limited English speaking students who participate in the district’s bilingual education program do not reach a parity of achievement with students who entered school already proficient in English at any time throughout the elementary grades and that since the district’s language program has failed to establish such parity, it cannot be deemed “appropriate action” under § 1703(f). Although this question was raised at the district court level, no findings were made on this claim. While under some circumstances it may be proper for a court to examine the achievement scores of students involved in a language remediation program in order to determine whether this group appears on the whole to attain parity of participation with other students, we do not think that such an inquiry is, as yet, appropriate with regard to RISD. Such an inquiry may become proper after the inadequacies in the implementation of the RISD’s program, which we have identified, have been corrected and the program has operated with *1015the benefit of these improvements for a period of time sufficient to expect meaningful results.14
To summarize, we affirm the district court’s conclusion that RISD’s bilingual education program is not violative of Title VI; however, we reverse the district court’s judgment with respect to the other issues presented on appeal and we remand these issues for further proceedings not inconsistent with this opinion. Specifically, on remand, the district court is to inquire into the history of the RISD in order to determine whether, in the past, the district discriminated against Mexican-Americans, and then to consider whether the effects of any such past discrimination have been fully erased. The answers to these questions should, as we have noted in this opinion, illuminate the proper framework for assessment of the merits of the plaintiffs’ claims that the ability grouping and employment practices of RISD are tainted by unlawful discrimination. If the court finds that the current record is lacking in evidence necessary to its determination of these questions, it may reopen the record and invite the parties to produce additional evidence.
The question of the legality of the district’s language remediation program under 20 U.S.C. § 1703(f) is distinct from the ability grouping and teacher discrimination issues. Because an effective language remediation program is essential to the education of many students in Raymondville, we think it imperative that the district court, as soon as possible following the issuance of our mandate, conduct a hearing to identify the precise causes of the language deficiencies affecting some of the RISD teachers and to establish a time table for the parties to follow in devising and implementing a program to alleviate these deficiencies. The district court should also assure that RISD takes whatever steps are necessary to acquire validated Spanish language achiev'ement tests for administration to students in the bilingual program at an appropriate time during the 1981-82 academic year.
AFFIRMED in part, REVERSED in part and REMANDED.
. The pleadings in this case also contained an allegation that the school district had administered the extracurricular programs of its schools with the purpose and effect of denying Mexican-American students an equal opportunity to participate in such activities. The record reveals no evidence on this issue and plaintiffs have not reasserted this claim on appeal.
. In Burdine, the Supreme Court elaborated upon the basic allocation of the burdens and order of presentation of proof in a Title VII case alleging discriminatory treatment which it had enumerated in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). The Court clarified the defendant’s burden of rebuttal by describing it as follows:
The burden that shifts to the defendant, therefore, is to rebut the presumption of discrimination by producing evidence that the plaintiff was rejected, or someone else was preferred, for a legitimate, nondiscriminatory reason. The defendant need not persuade the court that it was actually motivated by the proffered reasons.... It is sufficient if the defendant’s evidence raises a genuine issue of fact as to whether it discriminated against the plaintiff. To accomplish this, the defendant must clearly set forth, through the introduction of admissible evidence, the reasons for the plaintiffs rejection.
-U.S. at-, 101 S.Ct. at 1094 (footnotes omitted).
Although the Court’s opinion in Burdine clearly disapproves of this circuit’s previous practice of requiring the defendant in a Title VII case to prove the existence of legitimate non-discriminatory reasons for a challenged employment decision by a preponderance of *995the evidence, we do not believe that Burdine affects the burden shifting device we have long employed in the distinctive context of claims alleging discrimination, whether in employment or other areas, by a school district with a history of unlawful segregation. The analysis we have employed in this latter type of case is not derived from McDonnell Douglas; even as we employed the now disapproved “preponderance of the evidence” requirement in most Title VII contexts, we distinguished the situation where a claim of employment discrimination was lodged against a school district which formerly operated a dual school system and imposed the even stiffer “clear and convincing” standard. Lee v. Conecuh County Board of Education, 634 F.2d 959 (5th Cir. 1981). The application of this standard under these circumstances, is consistent with the type of presumptions approved by the Supreme Court in Swann v. Charlotte-Meckienburg Board of Education, 402 U.S. 1, 91 S.Ct. 1267, 28 L.Ed.2d 554 (1971) (in school district which formerly operated segregated dual system, burden placed on district to establish that continued existence of some one-race schools is not the result of present or past discriminatory action by the district) and Keyes v. School District No. 1, Denver, Colo., 413 U.S. 189, 208, 93 S.Ct. 2686, 2697, 37 L.Ed.2d 548 (1973) (“finding of intentionally segregative school board actions in a meaningful portion of a school system ... creates a presumption that other segregated schooling within the system is not adventitious ... and shifts to these authorities the burden of proving that other segregated schools within the system are not also the result of intentionally segregative actions.”) We do not believe the Court in Burdine intended to affect the manner in which this court has applied a presumption similar to that recognized in Swann and Keyes, to place on school districts having a history of unlawful discrimination a more onerous burden of rebuttal in an employment discrimination case than is usually imposed on defendant in a Title VII case.
. The record contains evidence that although Raymondville has always operated only one secondary school facility, attended by both Anglo and Mexican-American students, there was historically, segregation of Mexican-American students at the elementary school level. From school board minutes it appears that in the early decades of this century RISD operated schools on only one campus. There were separate buildings or wings of buildings on this one site for the “Mexican School” and the “American School,” both of which provided instruction in the elementary grades, and the secondary school which housed junior high and high school students.
In 1947, overcrowding at the central campus prompted a proposal that RISD operate another elementary school at a different site in northwest Raymondville and to establish attendance zones for elementary students. This proposal met with organized and vocal opposition from the Mexican-American community. The League of United Latin-American Citizens petitioned the board to consider another location for the new school and complained that the proposed site coupled with the new attendance zone policy would result in the establishment of a school attended almost exclusively by Mexican-Americans. The school board nevertheless proceeded to open a school on the northwest Raymondville site. This school, known first as the San Jacinto school and later as the North Ward school, was housed in old military barracks. This school was closed and the L.C. Smith school was built on the same site in 1962. We note that although the northwest campus has apparently been a virtually all-Mexican-American school, it is not clear from the record that the main campus elementary school was ever exclusively, or even primarily, Anglo and it is certainly not so today. It is clear, however, that as a result of the manner in which attendance zones were defined, the Anglo students were concentrated at the main campus elementary school facilities. At that campus, Mexican-American students were apparently instructed in separate classes during the first three elementary grades in an effort to provide English language instruction; classrooms at the main elementary school were integrated beginning with the fourth grade. The record in this case does not contain evidence from which we can determine whether, despite this history, RISD has now fully remedied the effects of these practices and operates a unitary system.
. We assume that the segregation resulting from a language remediation program would be minimized to the greatest extent possible and that the programs would have as a goal the integration of the Spanish-speaking student into the English language classroom as soon as possible and thus that these programs would not result in segregation that would permeate all areas of the curriculum or all grade levels.
. Singleton v. Jackson Municipal Separate School District, 419 F.2d 1211 (5th Cir. 1970) which set forth the standard form desegregation order in this circuit, required, inter alia, that:
Staff members who work directly with children, and professional staff who work on the administrative level will be hired, assigned, promoted, paid, demoted, dismissed and otherwise treated without regard to race, color or national origin.
Id. at 1218.
. The district court’s failure to make findings regarding the history of RISD does not impair our review of the merits of plaintiffs claims that inadequacies of the district’s language remediation programs render it unlawful because this claim is premised only on Title VI and the EEOA. The plaintiffs in this case do not argue that the current English language *1005disabilities affecting some of the Mexican-American students in Raymondville are the product of past discrimination or that the district is obligated to provide bilingual education or other forms of language remediation as part of a remedy for past discrimination. Cf. United States v. State of Texas, 506 F.Supp. 405 (E.D. Tex. 1981).
. RISD’s program was apparently adopted in compliance with Tex.Ed.Code Ann. § 21.451 (Vernon 1980 Supp.) which required local school districts to provide bilingual programs for students in kindergarten through third grade. The Texas legislature, although requiring and funding bilingual education programs has, nevertheless, provided that English shall be the basic language of instruction in Texas’ public schools and that bilingual education may be employed “in those situations when such instruction is necessary to insure that [students acquire] reasonable efficiency in the English language so as not to be educationally disadvantaged.” Tex.Ed.Code Ann. § 21.109 (Vernon 1980 Supp.).
. We think § 1703(f) clearly imposes on an educational agency a duty to take appropriate action to remedy the language barriers of transfer students as well as the obstacles confronting students who begin their education under the auspices of that agency. However, the challenge presented by these transfer students clearly poses a distinctive and difficult problem. Transfer students may bring to their new school varying amounts of previous education in English or another language; a school district may enroll only a few transfer students or may have a rather large revolving population of transient or migrant students who transfer in and out of the system. Factors such as these may be relevant to a determination of whether a school’s language remediation program for such students is appropriate under § 1703(f). In this case, neither the pleadings nor the record in this case indicates that the *1006distinctive problems presented and confronted by these students were addressed with the care necessary to determine whether RISD was currently taking “appropriate action” to meet their needs. Therefore we shall express no opinion on this issue in this decision.
. In Pennhurst State School v. Halderman,U.S. -, 101 S.Ct. 1531, 67 L.Ed.2d 694 (1981), the Supreme Court was called upon to determine the meaning of § 6010(1) and (2) of the Developmental^ Disabled Assistance and Bill of Rights Act, 42 U.S.C. §§ 6001-6080, which stated in relevant part that:
Congress makes the following findings respecting the rights of persons with developmental disabilities:
(1) Persons with developmental disabilities have a right to appropriate treatment, services, and habilitation for such disabilities.
(2) The treatment, services, and habilitation for a person with developmental disabilities should be designed to maximize the developmental potential of the person and should be provided in the setting that is least restrictive of the person’s liberty.
(3) The Federal Government and the States both have an obligation to assure that public funds are not provided to any institution ... that (A) does not provide treatment, services, and habilitation which are not appropriate to the needs of such person; or (B) does not meet the following minimum standards....
Id. at -, 101 S.Ct. at 1537. Plaintiffs in
Pennhurst urged, and the Court of Appeals had agreed, that this section imposed upon states an affirmative obligation to provide “appropriate treatment” for the disabled and created certain substantive rights in their favor and a private right of action to sue for protection of these rights. The Supreme Court disagreed. The Court, at the outset, analyzed the statute to determine whether Congress in enacting it had acted pursuant to § 5 of the fourteenth amendment or pursuant to the Spending Power and cautioned against implying a Congressional intent to act pursuant to § 5 of the fourteenth amendment, especially where such a construction would result in the imposition of affirmative obligations on the states. Id. at -, 101 S.Ct. 1538.
Although we are sensitive to the need for restraint recognized by the Court in Pennhurst, it is undisputed in this case, and indeed indisputable, that in enacting the EEOA Congress acted pursuant to the powers given it in § 5 of the fourteenth amendment. The general declaration of policy contained in § 1701 and § 1702 of the EEOA expresses Congress’ intent that the Act specify certain guarantees of equal opportunity and identify remedies for violations of these guarantees pursuant to its own powers under the fourteenth amendment without modifying or diminishing the authority of the courts to enforce the provisions of that amendment.
. The district court in its memorandum opinion observes that there was “almost total disagreement amongst the witnesses, experts and lay persons, as to the benefits of bilingual education and as to the proper method of implementing a bilingual education program if determined to be in the best interests of the students.” Insofar as this statement was intended to suggest that there was uncertainty and disagreement manifested in the record about the effectiveness of the bilingual education program currently conducted in Raymondville, it is certainly correct. However, this statement should not be understood as suggesting that the record in this case presents a dispute about the value of bilingual education programs in general. The issue in this case was not the soundness or efficacy of bilingual education as an approach to language remediation, but rather the adequacy of the actual program implemented by RISD.
. The only reference to the district’s in-service teacher training program in the district court’s memorandum opinion was an observation that RISD “is training non-Spanish speaking teachers in accordance with a State-administered program.” This observation does not constitute a finding that this program was an adequate one, nor a finding that RISD teachers who complete the program are adequately prepared to be effective teachers in a bilingual classroom.
. On remand, the district court should, of course, consider any improvements which may have been effected in RISD’s in-service training program during the pendency of this litigation.
. Dr. Jose Cardenas, plaintiffs principal expert witness on the subject of bilingual education, testified that he had no objection to the tests recommended by TEA for use in assessing students entering a bilingual education program. R. at 291. Mr. Inez Ibarra, employed as *1014principal of the L. C. Smith School at the time of trial in this case and who had previously served as bilingual education supervisor for RISD, testified that RISD had adopted, for use beginning in the academic year 1978-79, the Powell Test for language placement which was “on top of the list” approved by TEA. R. at 366.
. We note also, that even in a case where inquiry into the results of a program is timely, achievement test scores of students should not be considered the only definitive measure of a program’s effectiveness in remedying language barriers. Low test scores may well reflect many obstacles to learning other than language. We have no doubt that the process of delineating the causes of differences in performance among students may well be a complicated one.