The READ Institute

What Does The National Research Council Study
Tell Us About Educating Language Minority Children?

by Charles L. Glenn 1.

Introduction

The National Research Council (NRC) recently published its report on the status of bilingual education research and the direction of future studies. The READ Institute commissioned an analysis of this critical document by Professor Charles L. Glenn of Boston University who serves on the READ Academic Panel.

The intent of the NRC study is to report on what we know about educating limited-English students at present—29 years after the first federal law on bilingual education was enacted—and to suggest further research to improve school programs. The authors of this study claim that we do not know very much about how bilingual children learn or how best to teach them, suggesting that many more years of research and much more money is needed to investigate these topics.

Glenn points out what this study actually accomplishes. In its honest appraisal of the field, this study totally demolishes the myths on which bilingual education is based. While the NRC study treads very gently around each one of the following conclusions, it in fact directly contradicts what has been bilingual education dogma for years:

• There is no evidence that there will be long-term advantages or disadvantages to teaching limited-English students in the native language.

• Teaching children to read in English first, and not in the native language, does not have any negative consequences.

• Emphasizing cultural and ethnic differences in the classroom is counterproductive—it leads to stereo-typing, reinforces the differences from majority children, and does not lead to better self-esteem for language minority children.

• There is no research support for the idea that teachers who are themselves members of minority groups are more effective than others who work with children from those same groups.

• The U.S. Department of Education’s management of bilingual education research has been a total failure: wasting hundreds of millions of dollars; using the research agenda for political purposes to justify a program that has not proven its worth; and not making its research available to the educators who could use it to improve their school programs.

Glenn concludes with several sensible ideas, briefly paraphrased here. It is time we trust what we do know about teaching poor children, regardless of ethnic background, taking language into account, and stop putting our faith in research on second language acquisition alone. We need to know more about how to overcome the underachievement of Latino children and why they are outperformed by children of other immigrant groups. "Bilingual education is not of itself the solution to the underachievement of any group of poor children."

Report

A few years ago, the distinguished historian of education Carl Kaestle of the University of Chicago wrote an article with the dishearteningly honest title "The Awful Reputation of Education Research.2" His own reputation needs no defense, nor does that of Stanford's Kenji Hakuta, but the sad fact is that the awful reputation of education research is too often deserved. Or so we are forced to conclude by reading the massive study put together by Hakuta, Kaestle, and others for the National Research Council.3

Funded by the U.S. Department of Education and by grants from the Spencer, Carnegie, Pew, MacArthur, Mellon and other foundations, the study reviews the research carried out over the past thirty years on the schooling of what it prefers to refer to as "English-language learners" rather than as "limited-English-proficient" students. It does so, as the name of the report indicates, in order to propose what research should be carried out in the future, and how it should be organized, implemented, and assessed.

The report is not intended, then, as a systematic account of what we can and should learn from the existing research about how to educate language-minority children effectively, but inevitably it will and should be used for that purpose by those who must make decisions now about the children who are in our schools now and about the teachers who are being trained or retrained now to serve those children. There has been too much mystification on the subject, too many assertions that "research proves" this or that when it does no such thing. The reader of this report's 400 pages should come away with at least a healthy agnosticism, and perhaps also with confidence in the few things that it seems we do know about what works.

But the "awful reputation" is unfortunately deserved. It turns out that something like $100 million has been spent on research and evaluation of bilingual education over the last thirty years, with remarkably little in the way of bankable results that have been applied to practice. What we do know after all these years is often not reflected in the practice of school programs, which roll on with a momentum and a logic all their own.

Let's look at what they report that we do know with some confidence. The report is divided into sections corresponding to domains of research and practice, which we will consider in turn.

Bilingualism and Second-language Learning

The NRC report provides an interesting overview of research on the relationship between the age at which a second language is studied and the development of proficiency in that language. In brief, older (say, teen-aged) learners are more efficient at mastering vocabulary and grammar, but are not so good at learning to pronounce a new language, and those who learn a language as an adult often never lose their accent, even if they come to use the new language exclusively. "The earlier the better" is not a good rule if efficient learning is the goal, though no harm seems to be done by early exposure to learning a second language and often--as the Canadian immersion studies show--it can result in greater language proficiency over the long run.

Early exposure to English in school for language-minority students is opposed by many bilingual-education advocates, however, often arguing in justification that children with a solid academic background in their first language acquired before coming to the United States tend to do better at academic work in English than do children without such a background. That seems like common sense, but it does not necessarily follow that the child's American school should pretend for five to seven years (the time often advocated) that the child is still in Honduras or Portugal. Nor would this be likely to work.

It seems likely that the main reason that some bilingual education advocates call for teaching language-minority children exclusively or mainly through their home language for the first few years of their schooling in the United States is to seek to retard what otherwise is likely to be a rapid shift to the use of English. Much as some of us would like to think that Americans could become as bilingual as, say, Swedes or Dutch, it seems very unlikely; one study cited "found that the amount of language loss occurring in the United States in a single generation would have taken about 350 years in other nations investigated" (page 40).

We often assume that the personality or the attitude of the language-minority student toward learning English are decisive factors, but the NRC report found little evidence for this in the research (page 39). Just about everyone with experience in schools believes that classrooms which stimulate students to use a second language are more likely to help them to learn it, but "few studies have tried to link classroom communication and the learning of linguistic features, and those that have done so have not been successful" (page 41).

The chapter concludes (as do each of the others that follow) that lots more research is needed, even though--or because--the research to date has produced so few usable results.

Cognitive Aspects of School Learning:
Literacy Development and Content Learning

There are, fortunately, a few glimmers of what we do know works for language-minority children. For example, we apparently know that direct instruction in phonics and other "processing" skills is more important for these children than it is for middle-class English monolingual children. While "whole language" methods have their place in stimulating language use, "many believe that [language-minority] children are at considerable risk in classrooms that provide only a whole-language environment with no direct reading instruction" (page 58). That's important to know.

Other findings are more pedestrian; for example, it is scarcely a breakthrough to learn that "skilled readers can tolerate a small proportion of unknown words in texts . . but if the proportion of unknown words is too high, comprehension is disrupted" (page 62).

The authors take a refreshingly agnostic position on one of the central articles of faith of bilingual-education advocates, that children must be taught to read first in the language which they speak at home. "It is clear," they note,

• that many children first learn to read in a second language without serious negative consequences. These include children in early-immersion, two-way, and English as a second language (ESL)-based programs in North America, as well as those in formerly colonial countries that have maintained the official language [of the colonizer] as the medium of instruction, immigrant children in Israel, children whose parents opt for elite international schools, and many others. . . The high literacy achievement of Spanish-speaking children in English-medium Success for All schools . . that feature carefully-designed direct literacy instruction suggests that even children from low-literacy homes can learn to read in a second language if the risk associated with poor instruction is eliminated (page 60).

Later in the report, indeed, the authors conclude candidly that "We do not yet know whether there will be long-term advantages or disadvantages to initial literacy instruction in the primary language versus English, given a very high-quality program of known effectiveness in both cases" (page 179). Only those familiar with the field can appreciate how heretical this concession is, and how unwelcome it will be to those who urge that it is essential to postpone exposure to reading in English until reading is solidly mastered in the home language. It is disheartening, however, to learn that,

• an insufficient attempt has been made to understand the cognitive processes underlying successful transfer of first-language literacy skill to the second language, the limitations on that transfer, the conditions that optimize positive and minimize negative transfer, or the differences between children who manage learning to read in a second language well and those who do not (page 72).

It is as though the celebrated dicta of Jim Cummins fifteen to twenty years ago, arguing on the basis of thin research evidence that transfer from first to second language would occur and indeed must occur if language-minority children were not to be permanently damaged in their cognitive development 5, had simply been accepted as sufficient evidence without further investigation of whether this is true, how it might actually occur and what can promote it. Or so it seems from the NRC account.

Perhaps the most interesting--though too brief and general--discussion in this section has to do with how the epistemological characteristics of different academic subjects may affect the ability of students with an uncertain proficiency in English to get their minds around a subject which draws upon domain-specific concepts and syntax (pages 67-68). The authors suggest that "integrating subject-specific terminology into language classes helps English-language learners better comprehend the subject matter" (page 68).

An important point follows from this. We know that "language interactions--questioning, expert explanations, discussions of alternative solutions, formulation of reasons for conclusions--contribute to the development of understanding of complex subject matter" in any curriculum domain. But what does this imply for how we should ensure that language-minority children can participate effectively in the more advanced levels of the curriculum? Should we hold them back from learning situations that may demand a higher level of functioning in English than they are capable of? Wouldn't that deny them even the opportunity to complete the academic program on a par with other students? Or should we seek to provide all of the more "complex subject matter" in their home language? But quite apart from the practical difficulty and expense of doing so, for how long should or could such "separate but equal" education persist? The longer it went on, surely, the more difficult it would become for students to make the switch to dealing with ever-more-complex material in English.

Finally, how can non-bilingual teachers learn to present academic content in a way that is accessible to language-learners without "watering it down" in a manner which serves neither their interests nor those of the monolingual English-speakers in the classroom? Some complex themes and materials can be presented adequately in ways which are syntactically more clear and comprehensible than is ordinarily the case (and perhaps for the benefit of all students!), but others cannot without a loss of significant nuance. Learning to listen for nuance and subtle connections is an important part of becoming an effective student, and one which should not be withheld from language minority students. Nor is it clear that simplification will always do the trick; "Abedi, for example, simplified items using syntactical structures only and was unsuccessful in increasing performance" by English-language learners (page 74).

The Social Context of School Learning

For example, we are familiar with the research of Stevenson and Stigler, suggesting that Asian parents and children believe that academic success is the result of hard work, with some children simply having to work harder or longer than others to achieve the same results, while Americans attribute success to ability and are wont to say complacently "I'm no good at math." But the NRC report tells us that,

• in the United States, however, these ethnic differences are eliminated or even reversed: second-generation Korean American children attribute success to ability more than do European American children . . . and high achievers across a variety of ethnic groups (African American, Latino, Indochinese American, and European American, all low-income) attribute their success to their high innate ability (page 92).

What is more, we learn, "Latino and Indochinese immigrant children . . . have similar perceptions of parental socialization strategies and similar theories of educational success and failure. Nonetheless, the Indochinese immigrants were found to perform better than the Latino children" (page 92). Why? And is there anything we can learn from that which could benefit the latter? Not even a speculation is offered, no doubt because of the sensitivity of the theme.

What about intergroup relations in the classroom? Contrary to the common practice of encouraging children to celebrate their ethnic distinctiveness, the NRC study finds that "to increase positive intergroup contact, the salience of group characteristics should be minimized, and a superordinate group with which students from different cultural and language groups can become identified should be constructed" (page 94).

In other words, well-meaning efforts to persuade the children in a class to identify how they differ "culturally" because of their differing ancestry are likely to be counter-productive. One of my daughters, aged 8, asked me the other evening for help with a homework assignment requiring her to identify her cultural group (she said that the teacher had told one classmate to describe her culture as "white American") and what its customs and values were; she was having difficulty coming up with any ways in which our family differs from those of her Latino and African-American classmates. I explained that "white" and "black" are not cultural categories, and that almost everyone in our society, including the most recent immigrants, share the same values and have the same hopes for their children. My daughter and I made a list of what we wear and eat and what holidays we celebrate and how we believe people should be treated, and agreed that José's family and Ariel's family were probably a lot like us. Differences are interesting, but it is what we have in common which is important.

Perhaps I may be permitted to quote the discussion on this point from my own recent book on the education of immigrant children in 12 countries (gratifyingly cited a number of times by the NRC report). I wrote:

• There is an unfortunate tendency toward stereotyping in attempts to adapt school programs to the supposed demands of cultural diversity. It is conventional wisdom, for example, that Latino pupils recoil from the competitive atmosphere that allegedly rules in American classrooms, and would do better if instruction were organized on a cooperative basis. Actual research on preferred styles, however, has in at least one case found that Anglo pupils were more cooperative than Latino pupils!6 Research has also failed to confirm the belief that Mexican-American teachers, parents, and children differ from their Anglo counterparts in how they perceive classroom interactions.7 Common sense and the research on styles of cognition suggest that there is far more difference within ethnic groups in how individuals learn or relate to school than there is between groups.

  Even when there are culturally-based differences in learning style, it is not clear what implications for instruction are appropriate. A study of American Indian children in school found that their participation styles were very different from those of non-Indian American children; however, Philips concluded, placing the Indian children in a separate class, while it would reduce that conflict, would prepare them poorly for participation in the wider society. "The point is not that one set of values or behaviors replaces the other, but that the children have access to both sets so that they can form from both their unique bicultural identity."8 Another study did not "find support for the common conclusion that adapting instruction to Native Americans' learning style will increase achievement."9 Earlier research found that "culturally relativistic" teachers, influenced by anthropological literature on culture and teaching for altruistic reasons, were ineffective with Indian children and reinforced their differences from majority children.10

  Attempts to make schools more supportive of minority pupils by incorporating elements of minority cultures run the risk of becoming patronizing; . . .. Too much "sensitivity" can mislead children about the real costs of making it in the dominant culture, a cost that cannot be wished away. . . .

  There can also be unanticipated consequences of school programs that stress cultural differences, including reinforcement of the belief, among children, that race and ethnicity are defining characteristics of fundamental importance.11

On this aspect of "social context," then, the NRC report could be a corrective to the assumption that we know decisively important things about individual children, and about how to educate them, when we know their ethnic background.

On the other hand, the report does not fall into the opposite mistake of a shallow optimism that we should "just put the children together and let them sort out the differences." Racial and ethnic assumptions are so deeply-rooted in our society (and probably in every society) that deliberate efforts must be made to foster equal status and positive interactions. This could involve, for example, creating situations in which the children who are members of minority groups can draw upon skills and information which they possess--such as a second language--to help a mixed group to complete a task (page 97).

The report points to the research showing how important what children learn at home is in preparing and sustaining them for successful school experience, but acknowledges that we "know almost nothing about [socialization practices] among Puerto Rican, Santo Domingan, Central American, Vietnamese, Cambodian, Haitian, or Cape Verdean families" (page 103). It's hard to believe, but that's what they tell us.

Student Assessment

At present, two conceptions of accountability for academic results are competing in American policy debates--and in Britain as well. One argues that we should set clear standards for what students should be able to know and to do at various stages in their schooling, and devise valid and reliable methods of determining whether they have done so. Consequences should follow from the results of these assessments, though whether the consequences should be for the students themselves, for their individual teachers, or for their schools and school systems is hotly debated.

Critics argue that it is simply unreasonable to expect that language-minority students in general (or, in a variation, poor children in general) will be able to perform as well on assessments conducted through the English language as other ("majority") students in general, and that holding them to the same standards will be unfair to them, to their teachers, and/or to their schools and systems, depending upon which menu of consequences is followed. Instead, a "value-added" approach is urged, by which the achievement of these students would be measured against either (a) their own earlier achievement or (b) their predicted achievement, based upon well-established ("norm-referenced") assumptions about students similar to them in ethnicity, language-dominance, family characteristics, or some other dimension.

Those who call for uniform standards retort that this is to assume that schools and personal effort are incapable ever of making up for the effects of student background, and that society will or should permanently make allowances for individuals from language-minority groups. Isn't this a form of tracking, with lower expectations for some because of their social class or ethnic identity? Making allowances, they argue, is the easy way out, but we should choose the difficult but essential course of challenging and helping language-minority children to achieve up to the common standards.

A variation on either of these approaches has been to propose that schools and districts be required to meet "opportunity to learn" standards before children are held accountable for their performance on assessments. In one form, this position insists that it's not fair to test children on material which their schools have not taught, and so the curriculum and the assessments should be aligned. On the other hand, because assessments must be standardized in some way if they are to be comparable and thus useful for accountability purposes, the alignment would seem to require that curriculum be standardized. This conflicts not only with our tradition of local control but also with the belief of many that increased autonomy at the school level is the key to improved performance.12

Another form of "opportunity to learn" is the contention that students and their teachers should not be held accountable for results until all of the conditions--primarily though not exclusively resources--are in place that are necessary for language-minority or other at-risk students to achieve to their maximum potential. Unfortunately, as the NRC report points out, "the research base for defining the most important and effective resources and conditions for English-language learners is very weak" (page 128).

That is a stunning conclusion, though passed over rather quickly. After all these years and all these studies, we reportedly still don't know what resources and what conditions make a difference for these students. What did those tens of millions of dollars go for? we want to ask. Are we still just flying by the seat of our pants in educating those millions of children? These are questions that come up again and again in reading this painfully-honest report.

The trend of national and state policy-making over the past several years is to insist upon the inclusion of language-minority children in programs to assess student performance; this is reflected in the Department of Education Organization Act of 1994, in Goals 2000, in the reauthorization of Title I and Title VII, and in the Reauthorization of the Office of Educational Research and Improvement, as well as in school reform legislation in Massachusetts and other states. It might seem obvious that we would want to know how well language-minority children are doing in school, but in fact for many years such accountability has been quietly resisted.

Several years ago I served on the Governor's Commission on Bilingual Education in Massachusetts; we found that there was essentially no data on how tens of thousands of students had done over 25 years in the state-mandated program, because the state office responsible had never developed a method of ensuring that accountability for results was built in. I was not surprised, since for nearly 20 years, as the state official responsible for urban education and equity, I had fought unsuccessfully to persuade my colleagues in the bilingual education office that knowing how students are doing is essential to holding schools accountable for serving them more effectively. I was told repeatedly--and the State Board was persuaded--that testing language minority students would damage their fragile self-esteem!

The NRC report summarizes the new requirements at the federal level as:

• Standards and assessments are to fully include English-language learners.
• Innovative ways of assessing student performance are encouraged, including modifications to existing instruments for English-language learners.
• Programs are to be evaluated with respect to whether they meet "challenging" performance standards, rather than on a normative or comparative basis.
• Evaluations are to be useful for program improvement as well as program accountability (page 132).

It sounds, at first blush, as though Congress has come down squarely on the side of uniform high expectations, but it seems likely that there will be a lot of wiggle-room in developing "innovative ways of assessing" and "modifications". For example, why not provide certain assessments in Spanish, to determine whether Latino children are making progress in various academic domains while acquiring proficiency in English? But how do we ensure that a test translated into Spanish still measures the same level of achievement, the experts ask, and in fact many language minority children are not that strong in their first language; "immigrant children may acquire a less-developed knowledge of grammar in their first language as a result of their limited exposure to use of that language in their new communities" (page 129). Indeed, "a frequent feature of immigrant bilingualism is that seemingly more difficult tasks may be performed better in the second than in the first language" (36). It's not clear, then, that testing children in their home language will be any more "fair" than doing so in English.

The NRC report provides an overview of the assessment challenges, but little assurance that solutions are within our grasp if we just make effective use of the results of the past thirty years of research. The authors raise many questions and propose many lines of future research, but offer very few answers.

Program Evaluation

The American Institutes for Research (1978) study "showed that students in bilingual education programs did not gain more than students not in such programs" (page 142).

The National Longitudinal Evaluation (1984, 1989) found, among other things, that "too heavy a concentration on any one aspect of the English-language learner's education can detract from achievement in other areas," and that student achievement "is facilitated by different approaches, depending on student background factors". Not surprisingly, "proficiency in mathematics when tested in English seems to require proficiency in English," at least in the later grades (page 144).

The Immersion Study (1991, often referred to as the Ramirez study) found little difference in results among three models of instruction, though both those that stressed Spanish and those that stressed English seemed to do better than those that did neither and thus were incoherent. Reassuringly, but hardly a ringing vindication, "students in all three programs realized a growth in English-language and reading skills that was as rapid or more so than the growth that would be expected for these children had they not received any intervention" (page 145). In other words, at least the programs didn't do them any harm.

The National Research Council Report (1992) on the statistical approaches used in the latter two studies found that they "did not provide decisive evidence about the effectiveness of bilingual programs," but there was some evidence that "kindergarten and first grade students [who] received academic instruction in Spanish had higher achievement in reading than comparable students who received academic instruction in English" (pages 145-46).

Turning to studies that attempted to summarize and draw conclusions from the results of research, Baker and de Kanter (1981) found that "the case for the effectiveness of transitional bilingual education is so weak that exclusive reliance upon this instruction method is clearly not justified" (page 146).

Subsequent studies returned to the same research to attempt to squeeze out more definite conclusions. Willig (1985) concluded that there were some positive effects of bilingual instruction in all major academic areas. On the other hand, Rossell and Ross (1986) and Rossell and Baker (1996) found that "structured immersion," with a specially-trained teacher who is fluent in the native language of the students but teaches exclusively in English, is more effective than instruction through the native language (page 147). Finally, the U.S. General Accounting Office (1987) conducted a truly goofy study, asking ten experts on bilingual education what they thought about bilingual education, with predictable results.

The conclusion reached by the current NRC report seems entirely sensible:

• we see little value in conducting evaluations to determine which type of program is best. First, the key issue is not finding a program that works for all children and all localities, but rather a set of program components that works for the children in the community of interest . .. . . we think it better to focus on components than on programs. As we argue later, successful bilingual and [English-language] immersion programs may contain many common elements (page 149).

Thus do they dismiss as poorly-conceived a research strategy that has consumed tens of millions of dollars in federal funds and, more seriously, has provided the pretext not to fix what was obviously broken in our programs for language-minority students until we had the result of yet another large-scale study to provide the definitive answer.

In one of their moments of unintentional comedy, the authors assure the reader that "it would be a serious mistake to say we have learned nothing from the enterprise. We see that five general lessons have been learned from the past 25 years of program evaluation:

1. Higher-quality program evaluations are needed.
2. Local evaluations need to be made more informative.
3. Theory-based interventions need to be created and evaluated.
4. We need to think in terms of components, not politically motivated labels.
5. A developmental model needs to be created for use in predicting the effects of program components on children in different environments" (page 151).

The non-researcher is stunned by the first two recommendations. Aren't they obvious, haven't they been obvious for 25 years? What have evaluators, and those who pay for evaluation, thought they were doing? As the later chapter on "issues related to the research infrastructure" documents painfully, management by the Department of Education of the tens of millions of dollars of program evaluation mandated by Congress has been a total failure. Either those administering the program have been incompetent, or (more likely) they have not taken seriously whether bilingual education was producing solid academic results. Bilingual education has apparently been considered a self-evident good, not as simply a means to the higher good of enabling language minority students to flourish and achieve.

The third recommendation is also obvious, and the fact that it is felt necessary to make it in 1997 is disturbing. "Programs should be designed so they are consistent with what is known about basic learning processes" (page 155). Yes indeed; that's why we try, in schools of education, to teach future teachers something about how children learn and, increasingly, about how proficiency is developed in a second language. But we would expect that those who have designed programs for language-minority students had taken care to make them consistent with what is known about how children learn.

My hunch, from studying how such programs are designed and adopted in this and other countries, is that the NRC report is quite correct in suggesting that no such care has been taken. Two competing theories have been in the air, each owing more to political than to educational developments: that language-minority children need to be taught for a number of years through their home language, or that children need to be exposed as early and consistently as possible to the language which they will use primarily for the rest of their lives. The advocates of the first position consider those of the second complicit in "linguicide" and cultural imperialism, while being considered in turn guilty of ethnic separatism and self-interest. The debate over what language to use has consumed so much energy that little has been left over for considering what ought to happen between teacher and student, in whatever language. Meanwhile, most teachers have muddled along as best they could with little guidance from solid knowledge about what works.

At this point, the authors of the NRC report drop in one of their little gems of good sense, which we could wish appeared on the cover rather than at page 158. Effective "bilingual" programs and effective "immersion" programs may be quite similar, characterized in both cases by

• Some native-language instruction, especially initially
• For most students, a relatively early phasing in of English instruction
• Teachers specially trained in instructing English-language learners

If they were willing (and I have no reason to believe they would not be) to add additional features, early and consistent integration with native speakers of English, and the expectation that they will achieve to the same high standard, we would be in complete agreement. This is consistent with the conclusions of the careful study of the same body of research by Christine Rossell and Keith Baker, itself cited by the NRC report.

Studies of School and Classroom Effectiveness

Unfortunately, the recent fashion has been to define as "effective" schools that are considered by observers to be "exemplary," whether or not there is concrete evidence in student outcomes. The charisma of a principal may thus lead to a school being selected for study, and the study may in turn conclude that a charismatic principal is what is most needed for a school to be effective.

• Independent measures of student achievement are not in the data set reported by most of these investigators. In schools or classrooms with large numbers of English-language learners, this is often the case because investigators could not find adequate student achievement data to verify the validity of the nominations (page 166).

As a result, it is difficult to have complete confidence in the prescriptions arising out of this literature, and even more difficult to know how the features of a successful school can be transferred to an unsuccessful school.

Once again, the authors of the NRC report offer some good common sense, suggesting that "one important way to raise teacher expectations is to raise student achievement by creating structures at a school and helping teachers acquire skills and knowledge needed to be more successful with students, rather than by exhorting teachers to raise their expectations" (page 175). That bears paraphrasing: don't tell teachers they should have high expectations for what their students will achieve, but show them instead how to help their students to achieve up to high expectations.

Common sense also triumphs in their approval of schools that "plan for the needs of newcomers (newly arrived students who immigrated to this country after the early elementary grades) and include in the design of their programs strategies to meet their needs" (177). Like so much else in the NRC report, this may seem obvious, but in fact it is all too common for such students to be simply dumped into a bilingual class of the appropriate age level, with students who are in their third or fourth year and should be doing their classwork almost exclusively in English. Teachers are then constrained to use the home language much of the time to accommodate the newcomers, with the result that the other children do not progress in English as they should. The presence of newcomers also tends to work against integration of bilingual and regular classes.

The common European model, by contrast, is to put all newcomers into a special reception class for one year or, in rare cases, for two, and then to integrate them into regular classes, with on-going extra support as needed. This usually includes the opportunity to continue, as an elective, to study the heritage language and culture. It should be noted (contrary to the widespread impression here) that most Western European educational systems have a higher proportion of immigrant children than is the case in the United States, and that in none of them are separate bilingual classes the norm.13

The NRC report reserves its warmest approval for the Success for All program developed by Robert Slavin and his associates, reporting that it "has significant and important effects on the achievement of English-language learners, regardless of whether they are in a primary language [i.e. 'bilingual'] or sheltered English program." They note also that,

• there is nothing in the Success for All literature indicating that cultural validation or cultural accommodation per se is an important element of the program or, indeed, that culture plays any direct role at all (aside from language). Success for All is an intensive, prescriptive, well-conceptualized program designed to help as many children as possible leave third grade reading at grade level (page 179).

In effect, Success for All operates on the premise that children are children and deserve to be taught in a challenging and effective way that enables them to succeed. As a result, culturally and organizationally "the program is no different for African American students in Baltimore than for Latinos in bilingual education or sheltered English programs in California or for Cambodians in an ESL program in Philadelphia" (page 180).

This approach, they note, "runs counter not only to much of the accepted wisdom in the school reform literature, but also to previous efforts to disseminate and replicate effective programs" (page 191) in its assumption that a model can in effect be plugged in successfully with different teachers and different types of students. If it turns out the Success for All does in fact represent such a replicable package, we may find the question of when to use home language and when to use English becoming a distinctly secondary consideration rather than the all-important symbolic issue it is today.

The authors report that a number of studies of instructional practices have shown, not surprisingly, that "effective teachers for English-language learners use explicit skills instruction for certain tasks, mostly (though not always) to help students acquire basic skills" (181).

The chapter ends with the usual catalog of issues on which more research is needed, some as all-encompassing as "to examine the effects of instructional interventions and social environments on the linguistic, social, and cognitive development of young children" (page 193) or "to determine the resources required for effective instruction of English-language learners in different contexts" and "the benefits and shortcomings of different improvement strategies" (page 197), and other issues as concrete as "to assess the effectiveness of newcomer programs" (page 194).

Preparation and Development of Teachers

These all seem worthy initiatives, but the report's account is entirely uncritical and seems to rely entirely upon self-reporting by the various programs. No evidence is presented of whether the teachers prepared by one program or another are more effective than other teachers, and in what ways. So we are not surprised that the section on research needs related to teacher training starts from the assumption that we have to learn everything. After all, as they note, teacher certification programs in general "have not been empirically validated". They may well ask whether the four exemplary programs described are "based on theory, empirical evidence, or expert judgment?" (page 268).

There seems, for example, to be no empirical support for the widely-held view (the basis of several of the programs described here) that teachers who are themselves members of minority groups are more effective working with children from those groups, or better able to adapt instruction to respond to cultural differences. Research is needed, they report, on what benefit teachers gain from "supervised internships, readings, and classroom discussion." And on "factors, including theoretical knowledge, needed to support teacher learning once in the classroom and ensure that what is learned in professional development is applied in the classroom." We need to ask whether teacher preparation should,

• differ for different levels of teachers (i.e., early education, primary, secondary), as well as for teachers with different training and experience (e.g., ESL, mainstream), and if so, how? How are the needs of novice and veteran teachers different, and how does staff development accommodate these differences? (page 269).

These questions are taken from the first three sections of research needs; seven more follow, and it would be wearisome to recount all the things we should, but don't, know according to the NRC report. Or is it possible that we do know something about what makes good teachers, but it is slippery to grasp what that is through research?

Estimating Population Parameters

The difficulty is knowing when a child's poor showing on a language assessment results from dominance in another language, and when he simply functions in English at the restricted level characteristic of millions of poor children, urban and rural, who are in no true sense bilingual (though they may be bi-dialectical to some degree). If the latter is the case, they are not going to be helped by instruction through a language in which they may be even less proficient than they are in English. We have tended to confuse the needs of the newly-arrived immigrant who is fluent and perhaps even well-schooled in a language other than English with the second- or third-generation member of an ethnic group which has largely switched to the use of English (at whatever level of academic sophistication) in the home.

European as well as American studies on language use in the home has found that, even among the first generation of immigrants, it is common for parents to use the heritage language between themselves consistently and with the children usually, but for the children to use the host country language among themselves consistently and to some extent with their parents. That is, the children may speak the heritage language seldom, though understanding it when spoken by their elders.14

Home Language of Mexicans in the United States, 1989-1990.15

In fact, as the NRC study points out, "there is no common definition of limited English proficiency that would yield commonality across studies, states, or even districts within given states" (page 299). One thing we do know, unfortunately, is that there is considerable academic failure among the Hispanic pupils who make up 73 percent of the language-minority total.

• Hispanic children (a group that includes English monolinguals as well as Spanish-speaking bilinguals), for example, score well below their non-Hispanic peers in reading throughout the elementary school years and end up on average about 4 years behind in secondary school (page 58).

Since black pupils also experience academic difficulties in disproportionate numbers, while immigrant children from non-Hispanic homes tend to do quite well in school, it might be worth asking whether there are factors other than unfamiliarity with English which play a decisive role. John Ogbu's hypothesis of essential differences between "voluntary" and "involuntary" minorities, or Jeff Howard's theory of why many black youth lack a sense of efficacy, and what can be done about that, might have more to offer to the effort to improve the achievement of Latino youth than more linguistic and cognitive research.

Social determinants and cultural factors in fact confound all attempts to show a clear connection between particular "treatments" and particular results, and yet the agenda of school-based research can by its nature take little account of factors in the wider society which are likely to have considerably more impact on how fast language acquisition and language shift occur than are any changes in classroom practices. Whether Haitian immigrant youth in general make their way through the American educational system successfully or not is likely to depend upon how they perceive the opportunities open to them in the wider society, and whether they identify with middle class or underclass African-American models, much more than with how many years of instruction they are given through Kreyol. Indeed, a case could be made that the greatest benefit to many immigrant children of being in a separate bilingual class has been to limit their exposure to the peer culture of inner-city youth.

The under-achievement of Hispanics in the United States and of Turks and Moroccans in northwestern Europe, I suggested in my recent book, may have less to do with language differences than with their status in the society and how they come to terms with that status. "Cultural differences" (a notably imprecise term) may work either for or against success. Portes and Zhou contrast Haitians in Miami with Sikhs in California, arguing that the cultural distance and lack of adaptation of the latter help to explain their relative success. Haitian immigrant youth in Miami, by contrast, are overly-quick to assimilate to the American models of behavior and attitude most available to them:

• Native-born black youth stereotype the Haitian youngsters as docile and subservient to whites, and make fun of the Haitians' French and Creole as well as their accents [in English]. As a result, second-generation Haitian children find themselves torn between conflicting ideas and values: to remain "Haitian," they must endure ostracism and continuing attacks in school; to become "American" (black American in this case), they must forgo their parents' dreams of making it in America through the preservation of ethnic solidarity and traditional values. . . . As the Haitian example illustrates, adopting the outlook and cultural ways of the native born does not necessarily represent the first step toward social and economic mobility. It may, in fact, lead to exactly the opposite. Meanwhile, immigrant youth who remain firmly ensconced in their ethnic communities may, by virtue of this fact, have a better chance for educational and economic mobility.16

Of course, the dysfunctional assimilation experienced by some Haitian youth is as much an indictment of a society that has not prevented the development of a demoralized underclass as it is of the choices made by young Haitians to identify with it. The readiness of immigrants to make a success of their new lives is, as Ogbu points out, generally admirable; the relevant question may be, whether the United States and other Western societies have lost their power to assimilate newcomers, whether they can recover the unspoken conviction and unquestioned habits that allow a civil society to function even as those who make it up change.17

Issues Related to the Research Infrastructure

• In retrospect, however, the past 20 years has not been a heyday for research on this topic. . . . Nor did the research systematically contribute to improvements in practice, partly because of problems with the research methodology--an overreliance on large-scale evaluations and effective/nominated schools research, as well as faulty and weak mechanisms for oversight of the research enterprise (page 377).

At the heart of the unsatisfactory progress in learning how to educate language-minority children most effectively seems to be the Office of Bilingual Education and Minority Language Affairs (OBEMLA). The problem seems to be three-fold: (1) lack of competence to manage a complex research agenda (page 315), (2) commitment to using research to bolster the case for the existing bilingual strategy, heavy with symbolic and political significance, rather than to learn what works, and (3) failure to make what is learned from the funded research available for improvement of school practice. OBEMLA's "capacity to manage research has been inconsistent [or consistently poor?] and a frequent source of controversy" (page 361). The authors report that "even within OBEMLA, officials who worked in Title VII [bilingual education] program administration were either unaware of research results produced in OBEMLA studies or unclear about their implications" (page 384).

Despite occasional gestures toward pulling together research results and making them available,

• instances of such synthesizing activities are outweighed in the historical record by complaints of inattention to results. There are two related problems: one is whether the agencies do anything with the research they have funded (read it, understand it, critique it, synthesize it, disseminate it); the second is whether researchers in the field have a sense of evidence being amassed, of new directions and questions coming from completed research, and of relatively secure knowledge accumulating. . . . The frequent complaints about ignored reports and lack of synthesis are symptomatic of the weak infrastructure of education research in general and of research on English-language learners in particular (pages 319-20).

The authors report that a budget analyst sought, in 1992, to review the 91 research or evaluation studies that had been carried out with $47 million from OBEMLA between 1980 and 1991, He could not find final reports for 40 of them, since all research files for the period 1978 to 1985 had been thrown away. Of those remaining, he judged that at best 29 might be useful in formulating policy (page 384). Whether in fact any were used to this end seems doubtful.

As a remedy, the authors propose the establishment of yet another coordinating committee to set the agenda for research and monitor what the various funding agencies and research centers do. They are concerned that "English-language learners are not incorporated into many studies that purport to be about all students," which would seem to suggest the desirability of more effective integration, while on the other hand they stress "the need to target resources deliberately and well toward understanding issues that are specifically about English-language learners" (page 332). The difficulty, as I know from my own experience in government, is that the existence of a unit concerned with a target population tends to function as a license for everyone else to ignore that population. The fact that tens of millions of dollars have been earmarked for research on bilingual education has no doubt encouraged others concerned with research to believe that they don't need to make special efforts to include the needs of language-minority children.

The report recommends that the Office of Bilingual Education and Minority Language Affairs play a leading role in setting and overseeing the research agenda. That seems wrongheaded, perpetuating the political domination of research in this field. How can OBEMLA be trusted as an unbiased messenger or build consensus when it itself is--and probably must be--an advocate for particular interests and approaches? That seems to me a strong reason to separate it completely from research, and from the evaluation of the programs which it funds. OBEMLA should concentrate upon encouraging states and school systems to implement "instruction that is grounded in basic knowledge" already available "about the linguistic, cognitive, and social development of language minority children" (page 4). The funding and coordination of research and evaluation should be completely separate and insulated from advocacy, ensuring that the needs of language-minority students are part of the overall educational research agenda. Research and evaluation should not reinforce by their separate development the segregated schooling that too of them now receive. As more is learned, the instructional models will undoubtedly change, but OBEMLA should not be responsible for the evaluation of the efforts for which it is the designated advocate. Too much is at stake.

The intensely ideological nature of the issue of language-minority education (alluded to here and there in the report) is not given enough weight in the discussion of the research system. The authors recommend, for example, that "researchers and agency officials with deep research knowledge" should make the decisions about what research is funded, but in fact few of these individuals possess the objectivity that would allow research to flourish that would threaten the existing paradigm. Kenji Hakuta, as chairman of the NRC study, may attribute to others his own well-deserved reputation for objectivity.

There does seem to be a fundamental inconsistency, in a report documenting at painful length how chaotic and unproductive is government oversight of the research effort in second-language acquisition, of recommending more government coordination.

In summary, the NRC report's review of the present state of knowledge, and of how we have managed to obtain such limited results with such a major expenditure of funding for research and evaluation, are useful, accurate (so far as I am in a position to judge), and significant, but the conclusions seem rather too conventional, too content to color between the lines. The reader who has never engaged in research on language acquisition is astonished to learn how many agencies and centers have a finger in the pie, and is tempted to believe that more might be accomplished if all of them got out of the business of managing research.

A few hundred or thousand researchers based in universities and school systems, with modest grants, and subject to the disciplines of peer review that are already well-established in professional circles, and communicating their results by the well-established channels which already exist independent of government, would be likely to advance the state of our knowledge far better than the present over-managed system which has produced such dismal results.

Priorities for Research

More importantly, these questions should have been asked and answered before two entire generations of language-minority students passed through our schools, from kindergarten through high school (those who persisted), with in all too many cases an inadequate education.

Some of what we apparently already know from research is so obvious, of course, that it seems scarcely worth finding out: "For example, Spanish-speaking children score higher on National Assessment of Educational Progress vocabulary items that have Spanish cognates (such a 'fiesta') than would be predicted by their overall performance on the assessment" (page 352). Whoopee! Does that mean we should include more such items, or eliminate them altogether lest these children have an unfair advantage?

But important and serious questions are also raised in this chapter, and we can only hope that many researchers will be inspired--and funded--to address them. Every advance in knowledge in this area is a good thing. It is not clear, however, that it is necessary--or feasible--to have research-based answers to these questions before choices are made about how to educate language-minority children. Too often in the past those choices have been deferred pending another study or another review of research. If it accomplishes nothing else, the NRC report should demonstrate conclusively that such delay is irresponsible. We should get on with what we already know, and part of what we know is that there is no magic bullet or one-size-fits-all solution that can be worked out by researchers and applied in every school.

The NRC report insists that we continue to pursue the chimera of determining, through research and evaluation, what the best "model" of education would be, even though they have shown that we know very little about how different children from different language groups and different social circumstances learn different skills and different subjects in different schools. They call for "a model of instruction that is grounded in basic knowledge about the linguistic, cognitive, and social development of language minority children. This model would be rich enough to suggest different programs for different types of students" (page 4). What sort of a coherent model could possibly take into account all of the human and institutional and curricula differences?

Despite this hesitation, the reader is tempted to say, Great, let's do it! Let's be sure that states and school systems are implementing good programs based upon solid research, and that they have adequate resources and good teachers to do so well. But that's not what the authors have in mind. "It would take time to formulate such a model," they continue, even though we have tons (literally) of government-funded research as well as experience with programs in thousands of schools; surely there must be someone smart and experienced enough to sit down and write up what we have already learned in an accessible way, so we can go ahead and start educating better?

When the model has been developed by "researchers from very different backgrounds" (and think how long it would take for them to agree!), the authors write,

• [it] would be implemented in a small number of settings . .. Throughout implementation, the process would be observed and described, and the implementation would be reworked. Once successful implementation has been demonstrated, the programs would be formally evaluated for outcomes. . . . hypotheses could be generated about the generalizability of the findings to new sites . .. At this point, the programs could be disseminated as promising, and experimentation in other local sites encouraged. Once the model had been validated across a wide range of settings . . . [and so forth] (page 4).

Years and years and years, no doubt, added to the thirty years during which such "pilot" efforts have been designed and implemented and evaluated countless times. And meanwhile linguistic minority children have been, by all accounts (though with some notable exceptions which should be replicated vigorously), deplorably educated. Basta, enough! Policy and practice should move ahead with what we already know.

Conclusions

• [p]riority should be given to legitimate research questions that are of strong interest to . . . educators, policymakers, and the public at large. . . . The major areas of concern common to these groups are program evaluation and accountability, the extent to which students are acquiring English and progressing academically, and the characteristics of programs that promote student development (page 7).

As so often, the reader wants to ask what in Heaven's name researchers and evaluators and those who set them to their tasks think they have been doing for the past 30 years if this has to be spelled out now? And why have we kept mandating a particular form of schooling for language-minority children if we know so little about whether it is working?

In fact, as the report correctly notes, federal and state policies supporting bilingual education were adopted in a "leap of faith," and "basic research did not help inform practice. Nor did program evaluation research . ." (page 24). And what research and evaluation did not do before this educational strategy was developed (bilingual education was mandated in Massachusetts in 1971, and is mandated still), they apparently have not done since.

I want to try to be fair. It is possible that, in order to make a strong case that a great deal of research should be undertaken (and funded by the government and foundations) in every conceivable direction, the authors of the NRC report have over-emphasized what we don't know from the enormous and very expensive amount of research and evaluation which has already been done. If that is their strategy, it may prove a dangerous one: non-researchers like me are tempted to conclude that more research would be throwing good money after bad, and that the effort should instead go into training classroom teachers in those techniques of instruction that we have reason to believe are effective.

My colleague Maria Brisk contends that we know a lot about good practices; that what we do not know about, and perhaps will never know about in view of the complexity and variation of all the factors involved, is what a complete model of good schooling for language-minority children would be. If she is right (and I greatly respect her experience, knowledge and integrity), we should be teaching those practices to every teacher and administrator, not just to those who are preparing to work in separate bilingual programs. Every teacher and every administrator needs to know about and understand the reasons for sound and proven practices in educating the language-minority children who are very likely to turn up in their schools.18

Her position is in fact consistent with the fine print of the NRC report, which concedes that "we need to move away from thinking about programs in such broad terms and instead see them as containing multiple components--features that are available to meet the differing needs of particular students" (page 158). Having made that concession, however, the report returns to the call for an overall developmental model which would "predict exact [!] nonlinear growth trajectories for the major abilities--not only the mean or typical trajectories, but also their variability" and so forth (page 158).

I am not competent to judge whether such a model is feasible or whether it will ever be developed and validated; perhaps it would be a very good thing if that were done. Those who work in education or pedagogy as an academic discipline have long sought to make the field an exact science comparable to the natural or even, more modestly, to the social sciences, with strong and reliable predictive power. If such a general theory of learning "linguistic, social, and cognitive skills" is ever developed, it should of course take language-minority children into account. May it happen, and soon.

But until that glad day comes, I think I know enough of our present policies and practices and of their results to conclude that the interests of language-minority children in our schools will be better served by principled and theory-based experimentation on effective schooling of poor children of whatever ethnic background, taking language into account in how they are assessed and taught, than by putting our faith in research on second-language acquisition. We already seem to know a fair amount, through research, about how to develop bilingualism among middle-class children, especially if their parents are bilingual, but we know very little about how to overcome the academic under-achievement of Latino youth, or why they are out-performed by youth from other immigrant groups.

The promising work of Robert Slavin and CRESPAR, the Center for Research in Educating Students Placed at Risk of Johns Hopkins University, is a good example of such an approach. Neither for nor against bilingual education, it seeks simply to prevent under-achievement by providing a coherent instructional program. This and other ways of organizing instruction for results can be adapted to include among its objectives the development of proficiency in two languages, those of the home and of the school, if that is the freely-chosen goal of parents. Much of the research discussed in the NRC report could help to achieve that goal, if ways can be found to translate it into practice.

What cannot be justified, however, is to continue substituting a preoccupation with the language of instruction for the essential concern that instruction be effective. Bilingual education, it has become clear, is not of itself a solution to the under-achievement of any group of poor children. It is time that those of us who support bilingual education--in my case, by sending five of my children to an inner-city bilingual school--insist upon honesty about its goals and its limits. Bilingual education is a way to teach children to be bilingual, but it possesses no magic answer to the challenge of educating children at risk. Bilingualism is a very good thing indeed, but what language-minority children need most is schools that expect and enable them to succeed through providing a demanding academic program, taught very well and without compromise, schools which respect the ways in which children differ but insist that these differences must not be barriers to equal opportunity.

Endnotes

1. Charles L. Glenn is a Professor of educational policy at Boston University and responsible for urban education and civil rights for the Massachusetts Department of Education from 1970 to 1991. He served as an invited reviewer of the draft of the National Research Council report.

2. Carl F. Kaestle, "The Awful Reputation of Education Research," Educational Researcher, January-February 1993, 23-31.

3. Diane August and Kenji Hakuta, editors, Improving Schooling for Language-Minority Children: A Research Agenda, Washington, DC: National Research Council, 1997.

4. Kenji Hakuta, Mirror of Language: The Debate on Bilingualism, New York: Basic Books, 1986.

5. For a characteristic statement, see Jim Cummins, "The Role of Primary Language Development in Promoting Educational Success for Language Minority Students," in Schooling and Language Minority Students, Sacramento: Office of Bilingual Education, California State Department of Education, 1982.

6. P. Concha, L. Garcia, & A. Perez, "Cooperation versus Competition: A Comparison of Anglo-American and Cuban-American Youngsters in Miami," Social Psychology, 95, 1975, 273-274.

7. I. S. Stewart, "Cultural Differences Between Anglos and Chicanos," Integrated Education, 13 (6) 1975, 21-23.

8. Quoted by Sandra Lee McKay, "Weighing Educational Alternatives," in McKay & Sau-ling Cynthia Wong, editors, Language Diversity: Problem or Resource? Cambridge: Newbury House, 1988, 350.

9. Kleinfeld & Nelson, 1988, quoted by A. Lynette Pitman, "The Debacle of Native American Education," typescript, 1995, 25.

10. J. Kleinfeld, "Positive Stereotyping: The Cultural Relativist in the Classroom," Human Organization, 34, 1975, 269-274.

11. Charles L. Glenn with Ester J. de Jong, Educating Immigrant Children: Schools and Language Minorities in Twelve Nations, New York: Garland, 1996, 614-15.

12. My own take on this is that a sensible solution would be to define more clearly than in the past what students should know and be able to do, but leave individual schools more free than in the past to decide how they will help students to meet that expectation.

13. See Glenn and de Jong, 1996, pages 403-501.

14. For a more extensive discussion, see chapter 5 "The Languages of Immigrants" of Glenn and de Jong, pages 251-334.

15. Adapted from Rodolfo O. De la Garza, A. Falcon, F.C. Garcia, & J. Garcia, J., "Mexican Immigrants, Mexican Americans, and American Political Culture," in Barry Edmonston & Jeffrey S. Passel, editors, Immigration and Ethnicity, Washington, D.C.: The Urban Institute, 1994, 237.

16. Alejandro Portes and Min Zhou, "Should Immigrants Assimilate?" The Public Interest, 116, Summer 1994, 20-21.

17. Micha Brumlik, "Was heißt Integration? Zur Semantik eines sozialen Problems," in Ahmet Bayaz, Mario Damolin & Heiko Ernst, editors, Integration: Anpassung an die Deutschen? Weinheim and Basel: Beltz Verlag, 1984, 81.

18. A good place to start would be with Brisk's forthcoming Bilingual Education: From Compensatory to Quality Education (Lawrence Erlbaum).