Hold Your Tongue

MONOLINGUALS CHERISH A NUMBER OF MYTHS about how a second language is acquired. On the one hand, the task seems terribly onerous. Americans who struggled to learn a foreign language in school recall the drudgery of memorizing vocabulary and grammar, not to mention the embarrassment of attempting actual communication. (Woe unto those who tried to use their high school French in Quebec!) As adults with little to show for the experience, they tend to despair of the whole idea of language learning. On the other hand, children make it look so effortless. They seem to "pick up" a strange tongue within a few weeks, chattering away with new playmates before their parents can utter a respectable sentence.

Although these perceptions reflect real phenomena, they are distorted by social prejudices. Many Americans conclude, for example, that the most effective way to learn a second language is to be "totally immersed" in it. Necessity seems the best motivator. Conversely, the option of relying on one's mother tongue appears to weaken the incentive to learn another. This immersion fetish – the idea that maximum exposure and maximum will are what count in language acquisition – inspires much of the skepticism surrounding bilingual education. According to this reasoning, if children are allowed to keep their life-preservers, they will never swim unassisted. 

"I got the total immersion method," claims Mark LaPorta, chairman of the Florida English Campaign. A second-generation American, he grew up speaking English, although "a big chunk of my people spoke Italian as the primary language. My father would rather have a conversation in Italian than in English." To uphold the tradition, LaPorta was sent to learn Italian from relatives in the old country. "My father put me on an airplane for southern Italy when I was five and my brother was four, kissed us on the forehead, and said, 'Speak or don't eat.' For thirty to forty days, there was no English, and we ate fine. My family on that side were all schoolteachers. We didn't have to sit down for an hour and make a class out of it. But I was dumped in and I learned good Italian." LaPorta is fortunate for the experience. Today, in contrast to most Euro-ethnics, he can communicate freely with his elders in their favored language. What child would not benefit from a similar opportunity? Yet he is mistaken to generalize lessons for non-English-speaking children in American schools. Too often, their experience is to be "dumped" into a strange environment, without relatives who can help them, and expected to use a poorly grasped language in learning to read and other challenging pursuits. Intimidation and confusion are hardly ideal conditions for acquiring English. Furthermore, the stakes are considerably higher. For minority students, falling behind means being labeled a slow learner (perhaps even "learning disabled"),<1> and it greatly increases their likelihood of dropping out. Life chances can hinge on school performance at an early age. 

Over the past generation, research has cast increasing doubt on the immersion fetish. Psycholinguistics – the study of language development and its role in cognitive growth – has strengthened the rationale for bilingual schooling. Yet science has been slow to penetrate the thicket of opinionated discourse about language. One reason is that the study of bilingualism is an evolving discipline whose findings are somewhat tentative. For example, researchers have discredited the notion that learning in two languages will confuse a child's mind, but have yet to establish what many suspect: that "balanced" bilingualism can be an intellectual advantage. Kenji Hakuta, a Stanford University psychologist, traces the history of this puzzle in his highly readable Mirror of Language: The Debate on Bilingualism.

Generally speaking, however, the results of such research have been poorly communicated to the public. As yet, we have no Carl Sagan of linguistics ("Billions and billions of sentences ...") Many otherwise informed Americans seem oblivious to the field's existence. Journalists, frequently unable to determine who the experts are, tend to spread more confusion than enlightenment. Because bilingual education is controversial, it is reported less as a pedagogical field than a political issue, with opposing "sides" given equal time. The result has been to lend credibility to critics whose expertise approximates that of the Flat Earth Society. "We think that native-language instruction holds kids back," says Kathryn Bricker, of U.S. English, and her words are broadcast to the millions via network television. Yet Bricker has no training or experience in the subject; her career has consisted of advocacy for English Only and immigration restrictions. By contrast, bona fide experts in second-language acquisition tend to have little patience with these simplistic debates or their media impresarios, and thus avoid them. This is unfortunate. 

IF SCIENCE IS OFTEN COUNTERINTUITIVE, psycholinguistics is especially so. Recent findings about bilingualism contradict many of our perceptions drawn from immediate experience. To wit: the detour of native-language instruction is often the best route to English acquisition. Accepting this kind of go-west-to-get-east idea means abandoning one worldview and embracing another. 

Largely unbeknownst to the American public, a conceptual revolution has taken place in this field since passage of the 1968 Bilingual Education Act. Two of its leaders are Stephen Krashen, of University of Southern California, and Jim Cummins, of the Ontario Institute for Studies in Education. Neither researcher began with any "ideological" agenda favoring multilingualism. Krashen was seeking more effective ways to teach English as a second language, while Cummins was unconvinced by the traditional rationale for native-language instruction. Initially, bilingual education was conceived as a temporary measure to cope with a "language mismatch" between home and school – a way to keep minority students from falling behind in other subjects while they learned English. But the researchers discovered a more significant benefit: a firm command of the first tongue facilitates the acquisition of a second. 

While that may sound paradoxical, it makes good sense when the underlying principles are revealed. Krashen explains that "humans acquire language in only one way – by understanding messages, or by receiving comprehensible input." That is, the more exposure to intelligible messages in a second language, the more second language will be acquired.<2> Without quality input, however, quantity is meaningless. Students learn very little English in a sink-or-swim classroom, where the teacher's words sound like undifferentiated noise. The brain does not process what it cannot understand; hence any benefit of second-language exposure is lost. 

Here is the first way in which bilingual education can promote English acquisition. It provides context that makes English more comprehensible. Krashen illustrates this principle with what he calls the "Paris argument": 

Pretend that you have just received, and accepted, an attractive job offer in Paris. Your French, however, is limited. (You had two years of French in high school and one semester in college, and it was quite a while ago.) Before your departure, the company that is hiring you will send you the following information, in English: What to do when you arrive in Paris, how to get to your hotel, where and how to find a place to live, where to shop, what kinds of schools are available for your children, how French companies function (how people dress in the office, what time work starts and ends, etc.), and specific information about the functioning of the company and your responsibilities. 

It would be very useful to get this information right away in English, rather than getting it gradually as you acquire French. If you get it right away, the world around you will be much more comprehensible, and you will thus acquire French more quickly. Anyone who agrees with this, in my opinion, agrees with the philosophy underlying bilingual education.

Two reading experts, Frank Smith and Kenneth Goodman, have postulated that acquiring literacy is analogous to acquiring language: "We learn to read by reading," by making sense of print. Extending this logic, Krashen reasons that "it will be much easier to learn to read in a language one already knows, since written material in that language will be more comprehensible." If anything defies common sense, it is teaching children to read in English before they have acquired English, needlessly complicating the task. Better to build on their strengths rather than their weaknesses. Literacy, mastered in the first tongue, never needs to be relearned in a second. "Once you can read, you can read. This ability transfers to other languages that may be acquired," Krashen explains.<3> "Reading is the major source of our competence in vocabulary, spelling, writing style, and grammar." Print also expands the child's background knowledge and exposure to understandable messages, compounding the benefit. 

Besides comprehensible input, there is another variable that affects the brain's response to second-language stimuli. Krashen calls it the affective filter, a term for all the subjective barriers – high anxiety in a new environment, low motivation to learn the language, fear of sounding foolish – that can keep comprehensible input from reaching our "language acquisition device." Simply put, the right attitude is essential. Smith suggests that it involves two things: an expectation of success and a desire to join "the club" of those who speak that language. Defeatism almost ensures that one will fail, and so do feelings of estrangement (whether condescension or inferiority) toward speakers of the second tongue. Among adults, these are common obstacles. Here is where young children sometimes have an advantage, with their prepubescent lack of self-consciousness and their eagerness to make new friends. More open and motivated than their elders, they tend to pick up simple conversational skills more easily. 

Yet this is not always the case. A positive attitude can be destroyed by insensitive educators. To devalue a minority child's language is to devalue the child – at least, that's how it feels on the receiving end. The longtime policy of punishing Chicano students for speaking Spanish is an obvious example. While such practices are now frowned upon, more subtle stigmas remain. Children are quick to read the messages in adult behavior, such as a preference for English on ceremonial occasions or a failure to stock the school library with books in Chinese. The "early-exit" approach to bilingual education, with its haste to push children into all-English tracks, may have a similar effect. Whatever the cause, minority students frequently exhibit an alienation from both worlds. Jim Cummins calls it bicultural ambivalence: hostility toward the dominant culture and shame toward one's own. Though the idea of enhancing students' self-esteem has been much ridiculed of late, it is especially germane to the problem of English acquisition. If Cummins and Krashen are correct, a negative sense of self can be a formidable obstacle to language learning. 

RICHARD RODRÍGUEZ STANDS OUT as an interesting exception. In Hunger of Memory, he describes a personal odyssey from "socially disadvantaged child" to Milton scholar to apostate Chicano with a disdain for bilingual education. Prolonging children's reliance on their native language is a trap, Rodríguez believes, insisting that Hispanics cannot have it both ways – they must choose between treasuring a Spanish identity and advancing in Anglo society. "Bilingual enthusiasts bespeak an easier world. They seek a linguistic solution to a social dilemma. They seem to want to believe that there is an easy way for the child to balance private and public." Whereas, for Rodríguez, there is no avoiding the psychological turmoil and sense of loss. "I was that child! I faced the stranger's English with pain and guilt and fear. Baptized to English in school, at first I felt myself drowning – the ugly sounds forced down my throat – until slowly, slowly (held in the tender grip of my teachers), suddenly the conviction took: English was my language to use." 

This is the Americanization myth internalized, the same either/or mentality that induced European immigrants to prevent their children from becoming bilingual in the pathetic hope of improving their English. But Rodríguez is a compelling writer when he recalls sitting mute and helpless in the classroom, as strange words buzzed around his head. It is hard to fault his parents for instituting, on the advice of concerned teachers, an English-only policy in the home, even though their own proficiency was limited:

At first, it seemed a kind of game. After dinner each night, the family gathered to practice "our" English.... Laughing, we would try to define words we could not pronounce. We played with strange English sounds, often overanglicizing our pronunciations. And we filled the smiling gaps of our sentences with familiar Spanish sounds.... 

Again and again in the days following, increasingly angry, I was obliged to hear my mother and father: "Speak to us en inglés." (Speak.) Only then did I determine to learn classroom English. Weeks after, it happened. I spoke out in a loud voice. And I did not think it remarkable that the entire class understood. That day, I moved very far from the disadvantaged child I had been days earlier. The belief, the calming assurance that I belonged in public, had at last taken hold....

But the special feeling of closeness at home was diminished by then.... The family's quiet was partly due to the fact that, as we children learned more and more English, we shared fewer and fewer words with our parents. Sentences needed to be spoken slowly when a child addressed his mother or father. (Often the parent wouldn't understand.) The child would need to repeat himself. (Still the parent misunderstood.) The young voice, frustrated, would end up saying, "Never mind" – the subject was closed. Dinners would be noisy with the clinking of knives and forks against dishes. My mother would smile softly between her remarks; my father at the other end of the table would chew and chew at his food, while he stared over the heads of his children. 

Yet, in another way, the experience of Richard Rodríguez is instructive. What so alarmed his teachers was "a typical silent period," Krashen says. It is "not pathological, but normal" for a non-English-speaking child to go for as long as six months without "producing" speech in the second language. As long as there is comprehensible input in English, English acquisition is taking place. Then one day the child suddenly starts showing off his or her new knowledge. This, rather than any coerced "determination to learn classroom English," probably accounts for the breakthrough that Rodríguez describes. In the meantime it is unlikely that he received much useful second-language exposure at home because his parents' English was quite poor. Nor is it clear that he got much in class. He probably benefited most from two other sources of English mentioned in Hunger of Memory: after-school tutoring by a kindly teacher and interaction with children in his predominantly Anglo neighborhood. Krashen suspects that "Rodríguez would have succeeded quite well without giving up Spanish at home, [a decision that] resulted only in estrangement from his family and did not contribute to his English-language acquisition." 

FOR WORRIED PARENTS AND TEACHERS, recent research on bilingualism provides some welcome news. "A sense of urgency in introducing English to non-English-speaking children and concern about postponing children's exit from bilingual programs" are unfounded, according to Kenji Hakuta and Catherine Snow, a colleague at the Harvard Graduate School of Education. Despite appearances, it turns out that adults and adolescents acquire languages more efficiently than children. Canadian studies have shown that one year of second-language study in the seventh grade is worth three years' in the first grade. The researchers add that "starting to speak English even as late as high school is no barrier to learning to speak it very well." These findings confirm Krashen's "input hypothesis." Older learners receive more comprehensible input because they understand more of what they hear and read, drawing on their greater knowledge and intellectual attainment. On the other hand, it makes sense to begin English instruction early, for the simple reason that learning a second language takes time – indeed, a good deal more time than is commonly realized. 

Conversely, there is no evidence that "native-language instruction holds kids back," as U.S. English claims. Quite the reverse. A premature transition to all-English classrooms seems to retard academic achievement. This is true not only for the sociocultural reasons noted above, but also for linguistic reasons. "Language is not a unitary skill, but a complex configuration of abilities," write Hakuta and Snow. "Language used for conversational purposes is quite different from language used for school learning." In other words, there are two types of proficiency. One is exemplified by the speech children use on the playground, interpersonal communication that is high in context and low in cognitive demands. The other is more complex, involving the ability to manipulate verbal symbols without the aid of physical gestures or oral feedback: the kind of language needed for abstract reasoning. Jim Cummins, who first elaborated and tested this distinction, concludes that children need much less time to develop conversational proficiency in a second language (one to two years) than academic proficiency (five to seven years). 

This creates a quandary for the schools. After a relatively brief exposure to English, many students sound fluent enough to make it in an all-English classroom. So there is often pressure to "mainstream" them at that point, although few have acquired the kind of English they need to keep up in class. Until the age of ten or eleven, all children are continuing to acquire complex grammatical structures in their first language, along with vocabulary, literacy, semantics, and a repertoire of linguistic styles appropriate for various occasions. Moreover, they are using these tools to cover academic terrain of increasing difficulty. As Cummins observes, English-speaking students "do not stand still waiting for the minority student[s] to catch up." So when the latter are placed in a regular classroom without fully developed English skills, they inevitably fall behind. 

Far more than academic speculation, the theories of Cummins and Krashen have shaped a number of innovative bilingual programs. One example is known as the Eastman model, after a school in East Los Angeles, in which instruction shifts gradually from Spanish to English, but students continue to develop skills in their native tongue through the sixth grade. Barrio children who once scored far below citywide norms now surpass them in language, reading, and mathematics. The curriculum has been so successful that in 1988 the L.A. school board approved a $20 million plan to replicate it throughout the district. Similar developmental approaches have also yielded remarkable gains. 

All that said, it would be misleading to claim that recent advances in understanding bilingualism have been "proven" beyond all doubt. Research continues and so does debate. Such is the case with all science. Hypotheses are tested, discarded, refined, and retested on the basis of empirical data. Knowledge gradually accumulates until one day the reigning theory is overthrown. Given the intuitive prejudices about language and the novelty of psycholinguistics (not to mention the hold of Americanization ideology), the continuing skirmishes over bilingual education are unsurprising. What is disconcerting, however, is the fixation of policymakers on a single, narrow issue: does bilingual education "work"? For the vast majority of researchers and practitioners in second- language acquisition, this is no longer a serious question. For critics, it remains the only question. 

AFTER A DECADE OF SNIPING, in 1985 came the frontal assault. Secretary of Education William Bennett denounced the Bilingual Education Act as "a failed path ... a bankrupt course," and a scandalous waste of the taxpayers' money: "After seventeen years of federal involvement, and after $1.7 billion of federal funding, we have no evidence that the children whom we sought to help – that the children who deserve our help – have benefited." According to Bennett's diagnosis, the program "had lost sight of the goal of [teaching] English" in favor of "enhancing students' knowledge of their native language and culture." Meanwhile, there were "alternative methods of special instruction using English," such as so-called "structured immersion" programs, that looked promising. Yet these nonbilingual approaches were denied federal funding under the Bilingual Education Act. The law should be changed, he argued, to give school districts "local flexibility" to choose instruction tailored to their students' needs. 

With its stress on anti-Big Government themes, Bennett's attack was transparently political. The fact that there was no federal "mandate" for bilingual education got lost in the demagoguery. Competition for grants had always been stiff because, in a normal year, funding was available to serve only about 5 to 10 percent of the nation's eligible students. But methodological requirements were minimal; schools merely had to make some use of the students' native language to qualify for grants under the Bilingual Education Act. If districts wanted subsidies for all-English approaches, they could apply under other federal programs (e.g., Chapter 1, emergency immigrant assistance, or migrant education, to name a few). 

Nevertheless, the Secretary's call for "flexibility" sounded reasonable when combined with his argument that research on the effectiveness of bilingual instruction remained inconclusive. In some studies, he asserted, "the mandated method ... seemed to be no better than [a regular classroom] without any special help." He charged that "too many children have failed to become fluent in English," while the Hispanic dropout rate remained "tragically as high now as it was twenty years ago." As long as no one knew what worked, Bennett reasoned, schools should be encouraged to experiment with alternative, English-only programs.<4> 

How can this reading of the research be reconciled with the diametrically opposite findings of Cummins, Krashen, Hakuta, Snow et al.? Is someone hallucinating here? fabricating evidence? Or are the two sides talking about different studies? In fact, there are two types of research at issue: basic research on the cognitive process of second-language acquisition and evaluation research of actual programs in actual schools. There is no question that the former has provided stronger support for bilingual education than the latter. The clarity and consistency achieved in small, controlled experiments has yet to be matched in large-scale comparisons of student progress. 

Clearly, it is unfair to judge an approach where it is poorly executed, suffers from a shortage of qualified teachers, or faces conditions – crime, drugs, poverty, homelessness – that make any kind of learning a challenge. Proof of universal effectiveness, with every child in every school, is a standard that no other pedagogy is expected to meet. On the other hand, it is undeniable that students in many bilingual programs are still scoring below national norms, and this could mean there is something wrong with the concept. Perhaps the success stories are aberrations, a tribute to dedicated teachers and involved parents rather than to the value of native-language instruction. Hakuta and Snow put the question this way: either bilingual education is no better than English-only approaches or "the evaluation studies are doing a poor job of measuring" its superiority. The researchers make a convincing case for the latter hypothesis.

Everyone complains about the poor quality of evaluation research. One perennial problem has been limited support from the federal government. In the first decade, federal expenditures under the Bilingual Education Act totaled approximately $500 million, but less than one-half of one percent went for research. To make matters worse, early standards were slipshod. Evaluators often lacked expertise in linguistics or language education, and political pressures sometimes came into play, both for and against bilingual instruction. 

Furthermore, there are inherent difficulties in designing studies of bilingual programs. Educational evaluations will almost invariably show that children have learned something between fall and spring, but they might well have learned more (or less) in a different program. So an independent basis for comparison is essential. While it is possible to chart student progress against normal curves, which predict "average" rates of achievement, it is hard to generalize from such findings because there are many variables that can distort the outcome (e.g., an exceptionally bright group of children, the presence of books in their homes, a school principal's incompetence, drug shootouts in the neighborhood). The optimum approach is to study a comparable group of children – in socioeconomic status, language background, achievement level, and other traits – who are receiving an alternative treatment from teachers of equivalent abilities, ideally in the same school, over a period long enough to gauge program effects. Needless to say, a perfect match is rarely found. While there are statistical techniques to adjust for pre-existing differences between groups, these are imperfect, especially as variables multiply. 

Some of the methodological pitfalls are obvious in first large-scale evaluation of bilingual programs, conducted in 1975-76 by the American Institutes for Research. After reviewing the progress of 7,000 students in Spanish-English classrooms in 150 schools, then averaging the results, the study found no "consistent significant impact" for bilingual instruction. Hence Secretary Bennett's claim that it "seemed no better" than sink-or-swim. But one reason may have been a built-in bias that favored the comparison group. At least two-thirds of students in all-English classrooms had previously been enrolled in bilingual classrooms – apparently, many had "graduated" after being been deemed fluent in English – and they naturally tended to outperform LEP students who had yet to be mainstreamed. But the researchers made no attempt to compensate for this anomaly. In effect, the successes of bilingual education were held up as evidence of its failure. 

A 1981 "review of the literature" by the U.S. Department of Education was even cruder in its approach. Researchers Keith Baker and Adriana de Kanter sorted through more than three hundred evaluation studies and threw out all but twenty-eight as methodologically flawed. Reaching into this grab-bag of educational treatments, they added up the verdicts on bilingual education, pro and con, then took an average. Not surprisingly, their conclusions were a wash. Results from poorly designed programs canceled the results of well-designed programs. Moreover, it was never really clear what was being compared. 

For the critics of bilingual education, however, this remains the preferred type of research. Like Secretary Bennett, they assume that language of instruction, ipso facto, defines an educational "method." Programs that use two languages are lumped together as transitional bilingual education, while English-only programs are labeled immersion(if geared to students' level of proficiency) or submersion (if not). Other features of program design and execution – how languages are integrated in the curriculum, how children are assessed and reassigned, how teachers approach E.S.L. instruction – fade into insignificance. With the pedagogical details out of the way, quantitative analysis is then performed on a selected group of program evaluations. In head-to-head competition between bilingual education and its alternatives, the outcome is usually "no difference."<5> 

A leading practitioner of this type of research, Christine Rossell, of Boston University, sounds quite authoritative when she concludes: "In second-language learning, 29 percent of the studies show transitional bilingual education to be superior, 21 percent show it to be inferior, and 50 percent show it to be no different from submersion – doing nothing. Altogether, 71 percent show T.B.E. to be no different or worse than the supposedly discredited submersion technique."<6> There is only one problem with Rossell's procedure. Computer programmers call it gigo, or "garbage-in, garbage out." 

Research that relies on simplistic program labels is worse than useless: it offers no guidance to educators, while sowing confusion among policymakers. As we shall see, it also conspires in politicizing the debate, by encouraging participants to choose sides "for" or "against" bilingual education, immersion, and E.S.L. The pedagogical reality is that there are various ways to teach children in their native language, some better than others, and there is considerable overlap among bilingual, immersion, and E.S.L. methodologies. Some of the most successful approaches, such as the Eastman model, use all three. Some of the least successful are bilingual in name only; children get mostly E.S.L., with sporadic tutoring from bilingual aides.<7> A 1985 survey in California determined that students' first language was being used, on average, only 8 percent of the time in so-called "bilingual" classrooms; sometimes it was never used. On the other hand, some programs labeled "immersion" feature substantial amounts of native-language instruction. 

Further confusing the issue is the analogy often drawn with total immersion in Canada, a remarkably successful approach in teaching French to middle-class, anglophone children. Students start school entirely in the second language, with English introduced gradually. By the end of elementary school, most reach fluent (if not quite native-like) levels of French – at no cost to academic achievement or competence in English. From modest beginnings in an English-speaking suburb of Montreal, French immersion has spread throughout Canada in the past twenty years. By 1988, 224,000 students were enrolled, or about 5 percent of the elementary and secondary school population. Such programs dispel any doubts about whether it is possible for children to learn entirely in a second language, provided that instruction is made comprehensible. 

The question is whether an all-English immersion approach is advisable for linguistic minority children in the United States. Absolutely not, according to the researcher who pioneered Canadian immersion programs. Wallace Lambert, of McGill University, explains that because English is a high-status language in Canada, French immersion poses no threat to students' native-language development. Even in Quebec anglophone children have strong social and economic incentives to become proficient in English (not to mention help from well-educated parents). But minority children in the United States, who are more likely to live in poverty and to feel the stigma attached to their culture, have limited opportunity or motivation to develop their native-language skills outside of school. For these students, Lambert warns, English-only immersion tends to displace "the critical linguistic system associated with the development of basic concepts from infancy on." He calls this subtractive bilingualism, an approach that can leave children "in a psycholinguistic limbo where neither the home language nor English is useful as a tool of thought and expression." By contrast, the Canadian children are the beneficiaries of additive bilingualism; they learn French without giving up English and without falling behind academically. 

Lambert and his colleague Richard Tucker have formulated a simple guideline for teaching children in a bilingual environment: "Priority in early years of schooling should be given to the language ... least likely to be developed otherwise – in other words, the language most likely to be neglected." In this way, students can emerge proficient in both tongues, and their achievement in other subjects will not suffer; it may even be enhanced. 

By a happy accident, the Canadian immersion model can be adapted to the United States in a way that benefits both majority-language and minority-language students. Two-way bilingual education offers total immersion for English speakers learning Spanish, say, and developmental bilingual education for Spanish speakers learning English. The main difference is that children learn from each other as well as from the teacher. Only a few such programs exist, but they show promise in turning out fluent bilinguals who also do well in other subjects. And there is an added dividend for minority students. Instead of being treated as deprived children in need of remedial education, they enjoy the status of "peer tutors," encouraged to share valuable skills with English-speaking classmates. This can make an enormous difference in their attitudes toward school and toward themselves. 

In sum, bilingual educators are no longer groping in the dark. There are some remarkable approaches that are "working" just fine. What's more, an evolving theory of second-language acquisition has illuminated why these models are effective, an essential step in replicating them elsewhere. Stephen Krashen and Douglas Biber have compiled a volume entitled On Course, describing seven bilingual programs in California and analyzing their successes. While there are significant variations among these and other exemplary approaches – supporting the idea that no one method works best – all seem to share an important emphasis: they seek to cultivate, not replace, children's native-language abilities.

TWO THINGS ARE STRIKING about the academic opponents of bilingual education. First is their refusal to engage the pedagogical theory underlying bilingual instruction, other than to claim that it remains "unproven." Keith Baker spent a career, first at the Department of Education and later as a consultant for U.S. English, poring over evaluations of program effectiveness. At one point, he claimed to have reviewed more than 4,000 studies and concluded that "not one credible bilingual education program has ever been implemented in spite of all the time, money, and effort given over to it." Yet Baker never delved into the reasons why. It was only "common sense," he argued, "that English is learned by exposure to English, not by exposure to another language." This was all the theory he needed.<8>

Christine Rossell, while employing a similar modus operandi, at least elaborates a hypothesis to explain her view that bilingual education is inferior to all-English approaches. "Time on task – the amount of time spent learning a subject – is ... a good predictor of achievement" generally, she reasons. Immersion provides larger quantities of English exposure; therefore, it must be a superior way to teach English. Yet she ignores the competing hypothesis: that language acquisition is a special case of learning, which depends on quality, comprehensible exposure, and that cognitive growth depends on native-language skills. Herbert J. Walberg, another proponent of time-on-task, was once asked to respond to the linguistic arguments for bilingual instruction. "I don't see where linguistics has much to do with learning theory," Walberg replied. He dismissed the work of Cummins and Krashen as lacking "empirical support." 

Time-on-task, however, remains no more than a debatable explanation of the evidence. Other researchers dispute its reliance on Canadian immersion, which is bilingual in both methods and goals, to claim the superiority of English-only approaches. What scientists normally do in this situation is to test their hypothesis. Rather than merely gathering data to support their own interpretation, they design and carry out experiments in which it could be disproven. Yet neither Rossell, Walberg, nor anyone else has subjected time-on-task to rigorous scrutiny as a theory of second-language acquisition. Despite their contemptuous statements about existing research ("wretchedly planned and executed," says Walberg), they seem disinclined to lead by example. 

This is the second striking thing about the critics. With few exceptions, they come from outside the fields of applied linguistics, language education, E.S.L., and bilingual education. Baker is a sociologist by training; Rossell a political scientist. Walberg, though an educational psychologist, shows little acquaintance with the literature on bilingualism. None of them can claim any practical expertise in the programs they are judging. 

Challenged on this point when appearing as an "expert witness" in a lawsuit involving the Berkeley, California, schools, Rossell made the novel argument that her lack of background was a virtue. "Applied linguistics is a field of fads," she testified. "Linguists tend to be ardent supporters of language-maintenance programs because they themselves love languages and, as a result, frequently succumb to wishful thinking. ... They think that everyone can be bilingual." Rossell went on to describe her own evaluation of Berkeley's programs for LEP children, which gave the district a clean bill of health. She found that students who received bilingual instruction in Berkeley did no better than those who received E.S.L. and individual tutoring; therefore, there was no need to expand bilingual programs. Under cross-examination, however, Rossell admitted that her conclusions were based on (1) three minutes of classroom observation, (2) brief conversations with teachers, and (3) a partial comparison of test results for about 20 percent of the district's LEP students. She had little idea of what methodologies were used in either program. As usual, Rossell merely compared programs labeled bilingual or E.S.L. This is not how educational experts conduct evaluations. 

Baker has distanced himself even further from researchers and practitioners in second-language acquisition, whom he describes as "a vested interest group." In 1989 testimony before the New York Regents, he dismissed the views of professional organizations like the New York State Association for Bilingual Education. "They are for bilingual education because bilingual education lines their pockets with state money," he charged. Walberg has also insinuated that anyone associated with the field is suspect "because their jobs depend on such programs. Getting information from such sources is like asking your barber if you need a haircut." Asked how a researcher like Stephen Krashen, who long specialized in E.S.L., could have a vested interest in bilingual education, Walberg responded with a sly smile: "Consulting fees." 

As it happens, the critics have been less than reticent about seizing such opportunities for themselves. Christine Rossell has turned her court testimony into something of a cottage industry, appearing in numerous cases on behalf of school districts being sued by minority parents. In Teresa P. v. Berkeley alone, she collected $129,049 in fees and expenses. Paid at the same rate – $125 an hour – Keith Baker allowed his pockets to be lined with $40,950, while Rosalie Porter, another detractor of the "bilingual bureaucracy," took home $12,937. Not bad for part-time consulting work (all three had other jobs at the time).<9> This sort of gravy train is unheard of among advocates for bilingual education. In the Berkeley case, the plaintiffs' expert witnesses – including Krashen, Hakuta, James Alatis of Georgetown University, and Courtney Cazden of Harvard – were reimbursed for travel expenses, but were paid nothing for their time. Researchers who evaluate bilingual programs or conduct teacher-training workshops do not get rich, either. 

No one has suggested that the academic critics of bilingual education are in it solely for the money. Why do they hurl this ad hominem charge at their opponents? Why condemn an entire profession as biased by a desire to secure funding and perpetuate a bureaucracy? One never hears such complaints about math teachers, science teachers, or foreign-language teachers, not to mention the researchers who evaluate them. (Nor are questions raised about whether these pedagogies "work," despite perennial concerns about U.S. student performance.) Either bilingual educators are uniquely venal, willing to sell out the interests of children to protect their own, or there is a double standard at work. Why is this field being singled out? Does it have anything to do with a belief that bilingual education is little more than a Hispanic "jobs program," in which professional opinion is contaminated by ethnic politics? 

THAT IS THE IMPLICATION of Rosalie Porter in her 1990 book, Forked Tongue: The Politics of Bilingual Education. Formerly director of bilingual and E.S.L. programs in Newton, Massachusetts, Porter is one of the few critics who has spent more than three minutes in a bilingual classroom. Having started school as a monolingual Italian speaker, she can claim personal acquaintance with the sink-or-swim method. Unlike Richard Rodríguez, she does not recommend this treatment for LEP students today. But based on her experience as a bilingual teacher, Porter believes that special help should consist largely of English instruction. In her book she recalls the frustration of having to teach children in Spanish when they were beginning to speak English. "I doubted that [there was a] magic transfer of reading skills from Spanish to English. ... I decided, quite on my own, and based on this firsthand experience, to devote most of the teaching time to intensive work on English-language skills." Porter's thinking developed further along these lines as an administrator. By the early 1980s, she began to espouse "a flexible approach to the education of limited-English children" – English-only alternatives to bilingual education – first in Newton and later in lobbying the Massachusetts legislature. 

Porter's rebellion against her school's instructional policy is hardly unique. Teachers often feel they know better – sometimes they do – and many follow their own instincts in the classroom (to the despair of researchers seeking program consistency). Bilingual educators are not immune to social pressures favoring a rapid transition to English. Some start out impatient to get native-language instruction "over with," then change their minds after learning more about second-language acquisition or after witnessing a successful bilingual approach. On the other hand, where children are making little progress, teachers have reason to be concerned. No doubt something is wrong with the program, and every effort should be made to diagnose and correct it. This necessitates a vigorous and uninhibited exchange of ideas, as Porter correctly points out. Yet that cause is hardly advanced by her own tactics. 

Forked Tongue devotes an entire chapter to detailing "corrupt, discriminatory practices" by "the bilingual establishment in Massachusetts." Several tales involve Porter's own alleged victimization by militant Hispanics.<10> (Obviously, one of her objectives is to settle scores with numerous enemies she has made during her career.) She makes repeated reference to a "political ideology" of bilingual educators that warps their pedagogical judgment. When her opponents are not "misdefining students' needs" or using "unfounded conclusions as a basis for decisions," their "entrenched biases" are limiting parental choices, "suppressing" reports on English-only alternatives, and "squelching legislative change." At the end of her diatribe, Porter still wonders at the "siege mentality" of bilingual educators and "the atmosphere of distrust" toward well-meaning liberals like herself. 

There is no question that polarization has distorted the pedagogical debate. Under fire from English Only zealots, researchers and practitioners are increasingly cautious about discussing the shortcomings of bilingual programs, because their comments are easily twisted. To take one recurrent example, Kenji Hakuta, in Mirror of Language, makes a forthright comment about the state of evaluation research: "An awkward tension blankets the lack of empirical demonstration of the success of bilingual education programs. Someone promised bacon, but it's not there." Hakuta is nevertheless an articulate advocate for nurturing native-language skills, and he makes a convincing case that the evaluation research is misleading. His opinions are based not on "political ideology," but on psycholinguistic research – his own and that of others. Hakuta is not uncritical of transitional bilingual education. (Nor should he be: researcher-cheerleaders provide little guidance for classroom teachers.) Yet he has come to regret his choice of words. The quotation has been cited repeatedly by Porter, Rossell, Baker, and assorted spokesmen for U.S. English to argue that bilingual education supporters have no proof for their claims. In the Berkeley case, it was used to undercut Hakuta's own testimony for the plaintiffs; in New York State, to oppose the Regents' plan; and on Capitol Hill, to advocate cuts in bilingual education spending. 

In this adversarial climate, professionals in bilingual education feel the pressure to act like politicians – to mince their words or risk harming the field. Debates over the benefits of transitional versus developmental approaches, which were quite lively in the 1970s, have died down now that all forms of native-language instruction are under attack. The slightest criticism of pedagogical practice can be used to indict the theory of bilingual instruction by enemies seeking to abolish it. A researcher's candor can have unforeseen effects in the policymaking process. Polemics like Forked Tongue serve as a kind of Miranda warning: anything you say can and will be used against you in a partisan struggle. The effect is to discourage free discussion about how to improve the schools. 

Yet what is most disingenuous about Porter's complaints of political bias is her own intimate association with U.S. English. While posturing as a "moderate," unaligned with either of "the extremist camps – the doctrinaire official-English supporters and the strident proponents of full bilingualism," she does not hesitate to accept financial support from English Only advocates. In 1991, Porter took over the Institute for Research in English Acquisition and Development (READ), founded by Keith Baker with a $62,000 grant from U.S. English. READ's board chairman is Robert Rossier, a longtime activist for U.S. English who has called bilingual education "the new Latin hustle." The operation continues to receive major support from the Laurel Foundation, the benefactor behind John Tanton's numerous projects; English Language Advocates, a California group that campaigned for Proposition 63; and U.S. English itself. 

"We are unaffiliated and entirely independent," Porter insists. "These grants are without strings. We are free to research and publish what we deem to be reliable work done by respected people." Porter declines to divulge the organization's budget, but says that "less than half" of its funding comes from U.S. English. She maintains that READ has endorsed no political agenda. "We have no relationship with U.S. English and no obligation except to put the money they give us to good use and to report back on how we spend it." Nevertheless, one academic adviser to READ resigned after learning of the U.S. English connection. Guadalupe Hamersma, a Chicago elementary school principal, has her own reservations about bilingual education, but believes that READ's funding sources have compromised its pretense of objectivity.<11>

IF ONLY SOME WISE and disinterested researcher would perform the study-to-end-all-studies, supply indisputable data to prove or disprove the rationale for bilingual education, and bring this unproductive debate to a close. Congress could finally settle the question "Does it work?" with an up-or-down vote. Educators could focus on educating and politicos could move on to other campaigns. Applied linguists could concentrate on improving instruction rather than fending off ideologues. Children could get on with learning. Unfortunately, such a study is unlikely to appear anytime soon. But there is some important new evidence to consider. 

A long-awaited report on bilingual and alternative programs, by Aguirre International under contract with the U.S. Department of Education, was released in February 1991. The study's design, which overcame many limitations of earlier evaluation research, has drawn praise from partisans on both sides. Its principal investigator was David Ramírez, a researcher associated with neither camp, and during data-gathering (1984-88), the contract's federal overseer was none other than Keith Baker. Commissioned in 1983 at a cost of $4.1 million, the study involved more than 2,000 LEP students, all Spanish speakers, enrolled in nine school districts in California, Florida, New Jersey, New York, and Texas. Its objective was to compare the relative merits of three instructional models: structured immersion in English, "early-exit" (transitional) bilingual education, and "late-exit" (developmental) bilingual education. Ramírez paid close attention to program detail, so that each label meant something consistent. He also took pains to achieve comparability among students, teachers, and classrooms, and to compensate for disparities in parents' income and education, students' years in the United States, teachers' experience and training, and so on. Rather than take a brief snapshot of student achievement, Ramírez followed its ups and downs over four years. Finally, he sought out the best examples of each approach, so as to minimize the problem of poor execution. 

On release, however, the Ramírez study brought less clarity than many had hoped. The Department of Education muddied the findings by declaring "the three most common bilingual education methods" [sic] to be "effective." Therefore, "school administrators can choose the method best suited to their students, confident that, if well implemented, it will reap positive results." Not much help for the perplexed. 

What the study actually found was more illuminating. First, there were few significant differences in achievement between immersion and early-exit programs, that is, between children taught almost exclusively in English and those taught mostly in English.<12> Second, children in late-exit programs, taught primarily in Spanish, had the most sustained growth in achievement. Third, students in all three groups took five or more years to acquire academic proficiency in English. The second and third findings – and, arguably, the first – are consistent with the hypotheses of Cummins (linguistic interdependence) and Krashen (comprehensible input). 

The Ramírez study's impact was weakened by a decision not to include late-exit programs in the direct comparisons of student progress.<13> Nevertheless, its vote of confidence for additive bilingualism was quite strong. In the two subtractive approaches, early-exit and immersion, children made reasonable gains in English language, reading, and math. But their growth slowed down by the third grade, leveling off parallel to (but well below) the "normal curves" of English-proficient children. By contrast, in the late-exit programs, academic growth accelerated over time. By the sixth grade, these students had yet to reach national norms, but showed promise of doing so. For Ramírez, the implications are clear: 

If your instructional objective is to help kids stay where they are – around the 25th percentile – then give them immersion or early exit and they'll keep their place in society. If your concern is to help kids catch up to the norming population, use more primary language. In the late-exit programs, they're growing faster in content areas and in English, too. It's really clear that you will not slow down a child's acquisition of English by providing large amounts of native- language instruction.

This last program, the highest-performing in the study, is located in one of the poorest sections of Brooklyn, New York. Enrollment is 99 percent black and Latino. Known as School District 19, by all external criteria it is the proverbial dead-end for minority children. "A bombed-out area," in the words of David Ramírez. Muggers and drug dealers control the streets. The district's headquarters, which also houses an elementary school, resembles a fortress with its chains and grates, metal shields over the doors, and guards to frisk each visitor for weapons. "Teachers are really remarkable to work under these conditions," the researcher adds. "I've never seen a district that faced as many challenges." Yet its bilingual program has distinguished itself as among the best in the United States. Ramírez credits Frank Arricale, the former superintendent, for resisting pressures to mainstream children quickly and building "one of the rare late-exit models." 

Carmen Dinos, a Brooklyn College professor who helped to coordinate the study in District 19, cites several reasons for the program's success. Foremost, she believes, is its "supportive atmosphere. Children have to be given an opportunity to grow at their own pace, instead of being cut off at grade 3," as early-exit programs do. Where learning in Spanish is encouraged, "they don't feel like outsiders." She also cites Arricale's success in assembling a well-trained staff, with sufficient numbers of fluent bilingual teachers. Then there is the transfer of Spanish proficiency itself. "Allowing children to continue in the bilingual strategy doesn't hurt their ability to learn English; it enhances it." A final factor, found in all the late-exit programs, is that Latino parents are more likely to help children with schoolwork in a language they understand. 

As for structured immersion, Ramírez offers very qualified support. If teachers are proficient in their students' native tongue, if they have been specially trained in oral language development, if they have mastered "sheltered English" techniques, immersion may be appropriate for some LEP children in the United States. "Quality immersion programs are better than doing nothing," he says. "But the danger is in lifting the cap [on funding for English-only alternatives to bilingual instruction] without any clear specifications. When most districts talk about immersion, they mean taking regular teachers and placing them in a bilingual classroom, looking for a way to exit kids as quickly as possible." In such treatments, students receive little more than sink-or-swim, with some E.S.L. on the side. "We know that doesn't work," Ramírez emphasizes. In a few situations – for example, where bilingual teachers are unavailable or where students speak numerous languages – an immersion approach "may be all we can do," he concedes. "But is it really what we want? It doesn't cost you any more to provide a late-exit program than an immersion program. It's just a philosophical or political question." 

If research on bilingualism has demonstrated anything over the past twenty years, it is that there are no short-cuts to English proficiency. Subtractive approaches, whether immersion or short-term bilingual, cannot speed up the process. Even under the best of circumstances, the average LEP child needs five years or more to complete the transition. A fixation on teaching English as quickly as possible fails to prepare students to compete on equal terms. As a group they remain "at risk,"
disadvantaged, stigmatized. On the other hand, additive approaches promise to break the cycle of underachievement. When their first language is cultivated along with English, students are equipped to develop normally. They enter the mainstream later, but with improved chances of success and with the added dividend of fluency in two languages. 

1. It is still commonplace to find LEP children classified as retarded on the basis of English-language I.Q. tests. A 1980 study in Texas found that Hispanic children were overrepresented in the learning-disabled category by 315 percent; Alba A. Ortiz and James R. Yates, "Incidence of Exceptionality among Hispanics: Implications for Manpower Training," NABE Journal 7, no. 3 (Spring 1983): 41-53.

2. The terms acquisition and learning are used interchangeably here. Krashen draws a distinction, however, to denote "two independent ways of developing ability in second languages. `Acquisition' is a subconscious process identical in all important ways to the process children utilize in acquiring their first language, while `learning' is a conscious process that results in `knowing about' language." He hypothesizes that while learned knowledge – memorized grammatical rules, for example – can serve a "monitor," or editing function, naturally acquired language is essential to a full range of communicative competencies; Stephen D. Krashen, The Input Hypothesis: Issues and Implications (London: Longman, 1985).

3. The "transfer" effect varies depending on "surface" differences between languages. According to Cummins, researchers "have reported highly significant correlations for written grammatical, discourse, and sociolinguistic skills in Portuguese and English" and slightly lower, yet still meaningful correlations between Japanese and English. Similar results have been documented for Spanish and English, Hebrew and English, Finnish and Swedish, and Turkish and German; Jim Cummins, Empowering Minority Students (Sacramento: California Association for Bilingual Education, 1989).

4. In 1988, Congress acceded in part to Bennett's wishes, diverting up to 25 percent of bilingual education funding to nonbilingual programs. 

5. One significant exception is a "meta-analysis" by Ann Willig, of the University of Texas at Austin, covering the same studies reviewed in the Baker-de Kanter report. When Willig adjusted statistically for differences in program, student, and teacher characteristics, as well as research methods – 183 variables in all – the end results were moderately favorable to bilingual education. More important, she determined that the better the study, the better the measured outcome of bilingual education: "In every instance where there did not appear to be crucial inequalities between experimental and comparison groups, children in the bilingual programs averaged higher than the comparison children" on achievement tests; "A Meta-Analysis of Selected Studies on the Effectiveness of Bilingual Education," Review of Educational Research 55, no. 3 (Fall 1985): 269-317.

6. Another way to look at this is that 79 percent show bilingual approaches to be as good or better than English-only approaches. This conclusion is more telling, Krashen argues, "since it means that the children have acquired just as much English with significantly less exposure to English. This confirms the underlying theory of bilingual education."

7. Or when the native language was used, it was used poorly. In an early method known as concurrent translation, teachers would simply repeat each idea in two languages. With everything translated on the spot, students had no incentive to pay attention during the English parts of the lesson and simply tuned out. Conversely, teachers had no reason to tailor the second language to their students' level. The end result was little comprehensible input in English. Thanks to advances in basic research, concurrent translation has been widely discredited. Theoretically grounded techniques are taking its place, in which subjects are taught either in the native tongue or in "sheltered English," but not in both.

8. In an August 1991 interview, Keith Baker did an about-face, contradicting his statements of the past decade. His latest conclusion: "Bilingual education works, but not the way Cummins says it does. It works in the first three years, but after that point it flips – children do better in all-English programs." Ever the empiricist, Baker attributed his conversion to new patterns he had discerned in students' test scores. Though he had yet to work out any theoretical explanation, he said the phenomena could be due to "mental fatigue" among LEP children in English-only classrooms.

9. Although the Berkeley school district faced a fiscal crisis at the time, it dipped into an emergency reserve fund for approximately $1.5 million to defend the lawsuit. By comparison, the district spent $480,000 each year to educate its 571 LEP students; all but $35-40,000 of this amount came from the state of California. In the end, Judge D. Lowell Jensen ruled for the school district. He cited testimony by Berkeley's consultants that "a good teacher is a good teacher," whether bilingual or not. This decision reversed a string of victories for the plaintiffs' public-interest law firm, Multicultural Training, Education, and Advocacy (META). Still, it was hardly the "landmark decision" that U.S. English has hailed. In declining to mandate bilingual education, Judge Jensen accepted META's legal reasoning, but disagreed that the facts of the case justified such an order.

10. As a "gringo" attending the University of Massachusetts at Amherst, she describes being brushed aside in favor of "unqualified latino students" (who allegedly received preferential treatment in grades and fellowships). As the director of an alternative program in Newton, she claims to have been the target of "unrelenting attack" by an evil state bureaucracy (which terminated financial aid after finding the district out of compliance with state law). As a bilingual education adviser to Secretary Bennett, she describes a political smear by Hispanic critics (who called her a Reagan appointee, of all things). Only her side is presented; where the truth lies in these incidents is anybody's guess.

11. READ's other board members are Abigail Thernstrom, of Boston University, and Esther Eisenhower, former director of E.S.L. programs for Fairfax County, Va. Its "academic advisory panel" includes Christine Rossell, Herbert Walberg, Charles Glenn of the Massachusetts Department of Education, and Richard Estrada, a syndicated columnist and former research director of the Federation for American Immigration Reform. Keith Baker was forced out as director of READ in January 1991; he and Porter both say they cannot comment on the reasons for his departure. Baker adds, however, that given Porter's "bias against bilingual education, I would look closely at any research she comes out with. She's just trying to prove that bilingual education doesn't work." After another internal dispute at READ, Porter resigned as director in October 1991. 

12. Children's achievement was compared on the basis of English-language tests in reading, language, and mathematics. Immersion programs were taught in English 94 to 98 percent of the time, with Spanish used only for clarification. Early-exit programs featured English instruction two-thirds of the time in kindergarten, three-fourths of the time in second grade, and almost exclusively by fourth grade. The one statistically significant difference came in kindergarten and first grade, when the early-exit bilingual students outperformed the immersion students in English reading, despite their lesser "time on task" in English. 

13. Late-exit programs used English less than 10 percent of the time in kindergarten, with the proportion rising to 33 percent in second grade and 60 percent in grades 4 through 6. Ramírez explains that, because of the controversial nature of the study, he decided on a conservative statistical approach. The school districts offering late-exit programs had no early-exit or immersion programs that met the study's criteria. So he ruled out direct comparisons across districts, although indirect comparisons were possible by plotting academic growth rates in all three programs against national norms.