Research and the Quest for 'Effective' Bilingual Methods
This spring, an elementary school in Dade County, Fla., will become the site of an experiment in bilingual education. After years of relying on an approach known as "transitional bilingual education," the district will test, in that school, a method known as "immersion."
Instead of teaching in the children's native language until they become more fluent--as is done in so-called transitional programs--the school will use English as the main medium of instruction from the start.
About the time the Miami school is making its shift, a group of researchers from California will be visiting bilingual-education programs around the country to see if they meet the criteria set for a new, $2-million, federally funded study intended to gauge the effectiveness of the immersion approach.
Unlike submersion, its now-illegal predecessor that is known also as the sink-or-swim method, structured immersion requires a bilingual teacher who knows how to teach language through content. "That's the heart of immersion," says a researcher who is working on the study.
As observers in the field of bilingual education are quick to note, the recent interest in immersion is the latest sign that, after 16 years of emphasis on one approach--transitional bilingual programs--change is in the wind. Last year, further intensifying the debate, the Reagan Administration proposed amendments to the Bilingual Education Act that would have allowed schools to use any "effective" approach in teaching children with limited proficiency in English.
In explaining their attempt to end the exclusive use of transitional programs, Reagan Administration officials cited the results of a study initiated by the Carter Administration. The study, criticized by some observers on methodological grounds, found insufficient evidence to justify the exclusive use of transitional programs.
The efforts to document--and perhaps increase--the number of acceptable ways to teach those children have brought the question of effectiveness squarely to the center of a controversy whose political and pedagogical threads are difficult to untangle.
The political issues include the appropriate role for the federal government in maintaining the language and culture of foreign-born children, as well as its role in promoting bilingualism for social and economic reasons. Some bilingual educators see this as vital, and requiring a reformulation of the goals of bilingual programs.
The pedagogical questions, which are not entirely separate, involve the issue of whether one approach produces greater gains in achievement and thus should be officially sanctioned for use in federally funded programs. Or, if there is no recipe, are some approaches better for some children?
In the 16-year history of federal involvement in bilingual education, studies have reached numerous conclusions, some of which conflict, on these questions. Anecdotal evidence supplied by instructors and researchers offers another view, which sometimes supports research data and sometimes belies it.
"The question is more complex than can be answered with a yes or no," says James Cummins, an associate professor at the Ontario Institute for Studies in Education in Toronto. Mr. Cummins, who is regarded as one of the foremost theorists in the field of second-language acquisition, expresses dismay at what he views as U.S. policymakers' failure to look below the surface of the issue.
"The assumptions underlying bilingual education are grossly oversimplified," Mr. Cummins says. "Policymakers just haven't looked at the data in any perceptive way. If they did, what they would find is that the causes of underachievement are complex. They are not the same for all language minorities, and not all language minorities underachieve."
What that suggests, Mr. Cummins and other researchers say, is that it is not realistic to plan policy on the assumption that there is one route to successful bilingual education. But at the same time, they point out, much research data exists that could help to shape policy in a more sophisticated way while improving bilingual education.
With the data, however, researchers offer numerous caveats. One central problem has been that since schools are legally required to help pupils with limited proficiency in English, it is illegal to have a "no treatment" control group.
Researchers also point out that the effects of poverty are difficult to separate from the effects of language deficiencies. And at least one recent study has indicated that the rates at which children learn lan-guage vary widely, and that such variations are due to the interplay of multiple factors.
Nomenclature has also been a problem in evaluation, bilingual educators acknowledge. Programs that have the same label often differ significantly in their instructional approach. Only by looking closely at individual programs can researchers really tell what approach is being used.
"There are more variations in programs [that carry the same label] than sometimes there are with programs with [a different] name. It's a major problem in evaluating a program," says David Dolson, a consultant with the California Department of Education's office of bilingual education.
A Meaningful Education
The goal of bilingual education--as established by federal law and reinforced by the Supreme Court's decision in Lau v. Nichols--is to teach English to children whose proficiency in the language is limited so that they are not "foreclosed" from receiving a meaningful education.
Since 1968, when the Bilingual Education Act (Title VII of the Elementary and Secondary Education Act) was passed, that goal has been pursued through one strategy, transitional bilingual education. Native-language instruction, as necessary, was required by the 1968 law and all subsequent revisions.
In the Lau decision, the Supreme Court left open the question of instructional methods. "Teaching English to the students of Chinese ancestry who do not speak the language is one choice," the Justices wrote. "Giving instruction to this group in Chinese is another. There may be others."
But the 1975 "remedies," written as guidelines for complying with the decision placed strong emphasis on tbe, requiring its use through the elementary grades.
Hence, when people talk about bilingual education, they are generally talking about tbe One component of a transitional program is usually a class in English-as-a-second-language, which teaches language arts in English.
"The goal is to learn English, to have students mainstreamed into regular classes. The long-term goal is to get an education," says Ramon Santiago, director of the Bilingual Education Service Center at Georgetown University in Washington, D.C., a federally funded bilingual-education service center. The broader educational goals for students with limited proficiency in English, Mr. Santiago says, are no different than those for any other children.
"The native language is utilized to provide academic content, in order to prevent academic retardation while the student is learning English," Mr. Santiago says.
The concern about "academic retardation" arises from the historically poor academic performance of language-minority students. A recent study of Title VII programs, conducted for the federal government by Development Associates Inc., a Virginia consulting firm, found that "most projects served Spanish-speaking students who were working below both national and local academic norms."
Other studies support that finding. "The Condition of Education for Hispanic Americans," a federal report issued in 1980, notes that "Hispanic children enroll in school at rates lower than those for non-Hispanic students, they fall behind their classmates in progressing through school, and their attrition rates are higher than those of non-Hispanic students."
An analysis of data from the 1982 National Assessment of Educational progress found that Hispanic 17-year-olds who were "other-language dominant" and those who used English as their dominant language scored an average of 9 percentage points below the national average. Black and white "other-language-dominant" pupils also scored below their English-speaking peers.
Research on the effectiveness of federal bilingual-education programs did not begin until the mid-1970's. Hence, federal officials acknowledge, the transitional approach with a strong native-language instruction component was adopted more on the premise that it made sense than on the basis of research data.
"Years ago, when [bilingual education] first started out, before the act, it was an assumption that if a youngster was Hispanic, came from a Hispanic background, and was doing poorly in school, that was attributable to his language skills," says Jesse Soriano, director of the Education Department's office of bilingual education and minority-languages affairs.
On that assumption, he says, educators decided that these children should be taught English as quickly as possible--a decision with which he does not disagree. "However, there seemed to be some com-mon-sense logic involved in saying, 'If that's what we want to do, then some of the initial instruction ought to take place in a youngster's native language."'
The emphasis on native-language instruction was fueled also by the interest in ethnic heritage that developed along with the broader civil-rights movements of the 1960's, researchers say.
Since 1968, there have been several major studies of the overall effectiveness of Title VII programs, as well as many evaluations of smaller programs. Studies of Canadian French-English immersion programs have also contributed significantly to the understanding of effectiveness, scholars in the field note.
The first large-scale evaluation of Title VII programs was conducted in the mid-1970's by the American Institutes for Research, a private company, under a contract with the federal government. The study had four objectives, including determining the impact of Title VII programs "in cognitive and affective domains."
The researchers looked at about 5,300 students enrolled in 38 Title VII programs in 117 schools and compared them with a group of 2,400 students from 50 schools who were not enrolled in Title VII programs. The Title VII pupils were classified by teachers according to their fluency in English; based on that classification, pupils were given one of several tests of achievement in English and Spanish language arts and mathematics computation in both languages. Pupils were tested in the fall and the spring of the same school year; students in grades 2 and 3 were retested one year after the first test.
The study found that in the area of English language arts, participation in an English-Spanish bilingual project "did not appear to produce gains in student achievement over and above what would have been expected had the students been assigned to a traditional classroom." The study also found that the Title VII pupils made gains in mathematics that were about equal to those made by their counterparts in regular classes. Participation in the bilingual projects did not affect students' attitudes toward school or related activity, the researchers said.
The next major evaluation of effectiveness research came in 1980. Shirley M. Hufstedler, then Secretary of Education, issued proposed regulations that would have mandated the use of transitional bilingual programs. But a White House task force questioned whether there was enough evidence on the effectiveness of tbe to justify requiring its use in federally funded programs and asked for an analysis.
The task of evaluating studies on transitional bilingual programs and other methods fell to Keith Baker and Adriana de Kanter, both of the Education Department. Reviewing the research literature on effectiveness, Mr. Baker and Ms. de Kanter used two criteria to decide which of the several hundred studies they reviewed were methodologically sound enough to include.
The two standards were: "true experi-ments in which students were randomly assigned to treatment or control groups, or [s]tudies using non-random assignment which controlled for possible pre-existing differences between the groups either by matching students in the treatment and comparison groups or through statistical procedures," according to a summary of their study.
They found 39 studies that qualified. Of those studies, 10 indicated that tbe had positive effects "pertaining to language performance." Twenty-six of the studies reviewed found no difference in second-language performance between students who participated in tbe and those who did not.
The review also found some research data--two studies--showing that tbe enhanced students' acquisition of mathematics skills. But it also found 15 studies that showed no difference between tbe students and others. Three studies reported that tbe had a negative effect on math achievement.
The researchers included studies of Canadian French-English immersion programs in their analysis. The success of those programs, in which English-background pupils are taught mostly in French, was central to their conclusion that immersion programs "seem generally to succeed quite well in both second-language and subject areas."
In their report, the authors urged the federal government to sponsor large studies of immersion to better gauge its effectiveness.
"The conclusion is straightforward," the authors contend in a recent summary of their work. "There is no justification in terms of educational effectiveness for the proposed August 1980 regulations which would have mandated tbe as the sole instructional method for language-minority children. tbe has had mixed success. Although it has worked in some settings, it has proved ineffective in others and has had negative effects in some places. Furthermore, alternative instructional methods have been found to succeed--even to be superior to tbe--in some schools."
"The conclusion of our report is that it's beyond argument that the schools can help these kids make remarkable strides," Mr. Baker said in a recent interview. "The problem is, nobody knows what the recipe is. And back in 1980, the office for civil rights was trying to write one recipe. And sometimes that recipe worked in some places for some kids, sometimes it didn't. There are other recipes that sometimes work with some kids in some places and sometimes don't. And we just haven't quite yet fully figured out what the key ingredients of success are.''
Both the air and Baker-de Kanter studies were criticized for their methodology. Both also illustrate the difficulties faced by those trying to evaluate the effectiveness of various bilingual programs.
"The air study started out with the premise that they were going to evaluate the Title VII program, the Spanish-English portion of the Title VII program," says Gilbert N. Garcia, manager of research and evaluation for the Education Department's office of bilingual education and minority languages affairs. "The main flaw there was that they assumed the Title VII program was a homogenous program, that if we find 150 grantees at the [local education agency] level, they're all going to be doing the same thing. That obviously was wrong."
Tracy Gray, a researcher with the Center for Applied Linguistics in Washington who worked for air during the study, points out other problems. In asking participating schools to nominate comparable schools without Title VII programs, the researchers found administrators reluctant to make suggestions, lest the control schools be found out of compliance with the Lau decision.
Ms. Gray noted also that the plan to retest students only five months after the original assessment was "a little weak." Nomenclature too proved troublesome--programs described as "bilingual" turned out to be English-as-a-second-language programs. "In many cases, we were comparing apples and oranges," she said. "You're talking about some very basic flaws before you even get out of the box. Even from the beginning, you're not talking about a fair sample. We never knew who the control kids were. Some may have been in bilingual programs at some point, although they weren't then."
The Baker-de Kanter report is also criticized by some bilingual educators. They point to methodological problems--the same assumption of homogeneity that sunk the air report--as well as to the researchers' failure to look more closely at the programs' characteristics. Some acknowledge also that the most alarming thing about it was its advocacy of immersion.
On that point, Mr. Cummins suggests, Mr. Baker and Ms. de Kanter failed to emphasize the strong bilingual nature of Canadian immersion programs.
Looking at the issue of effectiveness from a different perspective, researchers at the Far West Laboratory in San Francisco observed 58 bilingual classrooms that had been judged effective by teachers, administrators, parents, and students.
The study found five common characteristics of effective programs. Successful teachers gave students a clear picture of what they wanted done and how to accomplish tasks competently, according to a summary of the research by Courtney B. Cazden of Harvard University. They also "communicated high expectations for lep [limited-English proficiency] students and terms of learning, and a sense of efficacy in terms of their own ability to teach."
Effective programs were also characterized by the use of two languages--English was used about 60 percent of the time and the native language the remaining time, Ms. Cazden reports. During instruction that was mostly in English, teachers used the native language to clarify meaning.
The fourth characteristic was the integration of two goals: "... the accommodation of instructional language to listeners who are not yet fully proficient, and the provision of activities that are designed specifically for learning [the second language]," according to Ms. Cazden's summary, part of a paper on "Effective Instructional Practices in Bilingual Education" prepared recently for the National Institute of Education.
The fifth characteristic identified in the study was the incorporation of the students' home and community culture into the classroom. "Cultural referents in both verbal and nonverbal forms are used to communicate instructional and institutional demands; instruction is organized to build upon rules of discourse from the [first language] culture; and values and norms of the [first language] culture are respected equally with those of the school."
Today, bilingual-education programs face what Ms. Cazden describes in her paper as "pressures to increase the instructional use of English."
These pressures, she writes, "come indirectly from the culture at large, and more directly from the use of tests in English as the sole criterion by which students (and thereby teachers) are evaluated."
Opposing that pressure, some bilingual educators argue that native-language instruction should be used for all language-minority children.
But according to Mr. Cummins, neither extreme--for and against native-language instruction--survives intact when viewed in light of the available data.
"There are two major assumptions, or rationalizations, that underlie the policy debate in the United States," he says. "Those who oppose bilingual education argue in favor of a maximum-exposure model: the more [English] they get, the more they will learn. Obviously, that makes some sense.
"But the opposite argument that proponents use also makes a lot of intuitive sense: If there is a home-school language switch, and children are taught through English, obviously they can't learn in a language they don't understand." That hypothesis is known as the "linguistic mismatch" model, he explains.
"It's fairly obvious that both of those have a lot of folk wisdom supporting them," says Mr. Cummins, "but when you look at data, each of those assumptions is virtually refuted by all the data. Neither is correct in a theoretical sense. Although it probably is more difficult to learn in another language--and if you're not exposed, you won't learn--you can't generalize those assumptions into a policy argument because there are all kinds of data that refute strong forms [of the arguments.]"
What the U.S. research shows, according to Mr. Cummins, is that children taught mostly through Spanish are at least as successful in learning English as those who are taught in English.
"That refutes the maximum-exposure hypothesis, but linguistic mismatch doesn't stand up any better," Mr. Cummins says. "If [linguistic mismatch] were valid, and being educated in a second language led to academic problems, we would find children in French immersion programs having academic problems. They don't." Rather, studies have shown that children from English backgrounds taught mostly in French--their second language--do better in all academic areas, at no cost to their first-language skills, he notes..
"When you look at research findings of language-minority [instruction] or if you look at it from immersion, you get very consistent data in both situations," Mr. Cummins says. "What you find is that the amount of instruction in English does not in any way relate to how well children do in English. So you've got a situation in terms of what we know which contradicts the two conventional wisdoms that policymakers jump back and forth from."
Cutural Factors Key
Other researchers point out that although native-language instruction is not essential for all children, there are pedagogical reasons linked to both achievement and culture that support its use for some pupils.
One important premise behind bilingual education is that language-minority children need special help not only because they cannot speak English, but also because they are likely to experience other academic difficulties as well.
The theory is based on the premise cited by Mr. Soriano of the Education Department--that "linguistic mismatch" is the main problem facing Hispanic children. Mr. Soriano says that he now questions that premise. Mr. Cummins suggests that the question is more complicated than that--linguistic mismatch is a problem for some students, while others overcome it quickly.
"There isn't a generally accepted or clear picture of why some minority children experience difficulties," says Mr. Cummins. "There are a lot of hypotheses that are plausible. We can say there are a lot of factors: social, linguistic, psychological among them. It's the complex interaction that leads to underachievement. We can say that it's a lot more complex than just linguistic mismatch."
Oriental students, Mr. Cummins says, frequently are able to overcome the difficulties of hearing one language at home and another at school. Part of the explanation may lie in cultural factors. Chinese families place a high value on education, and the student who does well in school upholds and augments the family honor. Chinese children tend also to be very "adult oriented," Mr. Cummins notes, a characteristic that predisposes them to obey the teacher even when the instructions make little sense. "Those characteristics seem to help children do well even under difficult situations," he says.
"If you look at characteristics of children who have tended to fail academically--French-speaking children in Canada, Finnish children in Sweden, Hispanic children in the United States--one of the things that jumps out at you is that most of these minority groups are characterized by an ambivalence about their own standing and about the majority society," Mr. Cummins says.
The children know that they must learn English to succeed in the dominant culture, but have mixed feelings about doing so, he explains.
If such children are taught only in English, he and researchers argue, children sense that their background is being rejected because it is not worthwhile. "By using the child's first language as a legitimate means of instruction, school is conveying an important message--that it does not need to be cut off," says Mr. Cummins.
Learning In Two Languages
There are also sound--and overlapping--pedagogical reasons for using native-language instruction when teaching children a second language, researchers have found.
Stephen D. Krashen, a professor of linguistics at the University of Southern California, says programs that use native-language instruction are more effective if they include two elements--comprehensible input in English and solid subject-area instruction in the native language.
"We acquire language in one way: when we understand messages," Mr. Krashen says. "If I speak to you in French, and you understand, that's the best lesson. It's called comprehensible input. If you want to teach someone English, you have to give them comprehensible input."
"You make things comprehensible by gesture, by background information, by context," he explains. "The main point is we acquire language by paying attention to what it means. Language-teaching methods that give people lots of comprehensible input are much more effective than drill-and-grammar methods," if effectiveness is measured by how well the child can speak, read, and write in the second language.
Instruction in the two languages interacts to the benefit of the child, Mr. Krashen says.
"A child who knows subject matter will acquire more English in an English-taught class than a child who knows less subject matter," he says. "Let's say we have two 3rd graders, both with limited proficiency in English. One child understands math very well. The other does not. The child who understands math will do much better in an English-taught math class than the other child. He'll get more math as well as more English because it's more comprehensible. The child who doesn't know his math gets neither math nor English."
This, Mr. Krashen says, "explains why bilingual programs that combine solid subject-matter teaching in the first language and comprehensible input in English teach English as well, and usually better than all-day English programs."
This approach also helps the child develop the cognitive and academic skills needed to progress in school, researchers say.
"You're learning underlying academic skills that are not specific to one language," Mr. Cummins says. "A child who's developed reading skills in Spanish will be able to transfer many underlying aspects to reading in English."
Designing A Program
Researchers differ in what components they include in descriptions of good bilingual programs; they also point out that more than one approach is acceptable, providing it includes certain elements.
A good program that adheres to the principles of comprehensible input in English and subject-area instruction would move from great reliance on the native language to English-only instruction, Mr. Krashen says. For example:
A child who enters kindergarten knowing no English should be taught only art, physical education, and music with English-speaking children.
The child should get good English-as-a-second-language instruction, with an em-phasis on comprehensible input and "natural language use," not grammar drills. Natural language use includes the discussion in English of topics relevant to the child's life in school. At this stage, all academic subjects would be taught in the child's native language.
Next, the child would continue to take art, music, and physical education in the regular classroom, and would continue to get esl instruction. But at that point, the child could be put in a "sheltered" math class for students with limited proficiency in English, which would be taught in English. The child would continue to learn social studies and other academic subjects in his or her first language.
Within one or two years, the child would move deeper into English, going from a sheltered to a regular math class and from a native-language to a sheltered social-studies class, Mr. Krashen suggests.
In the fourth and final stage, the child would be instructed exclusively in English, but could take elective courses in the native language. "The whole program may take as long as six years," he explains. "If there's no English spoken on the playground, that's what you'd expect. If there is a lot of English spoken on the playground, it will take less time." Evidence to support this approach comes from studies of Canadian immersion programs, Mr. Krashen says. It takes students five or six years to learn French, but at the end of that time, their French is almost as fluent as that of a native speaker of French.
A key factor, he notes, is that "there's always a real message being transmitted."
Children should make the transition to English-only classes when teachers are confident that they have "sufficient proficiency to cope with content and basic skills," adds Jose Cardenas, executive director of Intercultural Development Research Association, a San Antonio organization.
Teachers should not make the decision simply on the basis of achievement-test scores, but on language proficiency and on their judgment that the child can handle English instruction.
As an optional component, Mr. Cardenas suggests, schools might want to help the child to maintain and develop further his or her skill in the first language. "It is not a key portion of the bilingual program, but it is a desirable social goal," he says.
"We have done evaluations of a lot of bilingual programs, and I've come to the conclusions that well organized and well implemented programs are extremely successful," Mr. Cardenas says, basing his statement on formal evaluations of about 30 programs and informal evaluations of several hundred.
Among the important variables, he suggests, are bilingual teachers, adequate time and materials, and administrative support. "When these are present, bilingual programs have been extremely successful," he says. "At this point, less than 10 percent of all bilingual programs in the U.S. meet the adequacy criteria that I've just described. Until you have adequacy of treatment, I don't think you can make any judgments on the outcomes."
In a number of studies now underway, researchers are working to sharpen their focus on the issue of effectiveness. The immersion study, which is being conducted by sra Technologies Inc., in San Francisco, received about 90 suggestions of possible immersion programs to evaluate, but only nine met the criteria established. After the researchers visit the sites, the number may shrink, project officials say.
The Education Department is also continuing its longitudinal impact study, which began during the 1983 fiscal year. The first phase involved gathering descriptions of the services received by limited-English-proficiency pupils in grades 1 through 5 and grouping the descriptions by various characteristics, including each program's reliance on native-language instruction, according to Mr. Garcia of the federal agency.
The second phase of the study will involve assessing program effectiveness. "The objective there is to track the sample of students for three years," Mr. Garcia says. The investigators plan to follow the progress of several thousand students who remain in bilingual programs as well as those who do not. Total funding for the study is about $6 million.
Federal officials acknowledge that they are interested in looking at a wide variety of programs in their quest for effective methods. They are also trying to profit from the flaws of past studies, and to look at characteristics rather than labels.
"There will be more and more studies," says Mr. Baker of the Education Department. "We will begin to get some better notion, some idea [of the important program characteristics.] And it may turn out that the key parts of the recipe don't very much resemble the way people have chopped up the world up to this point."