- Structured immersion—or sheltered English immersion—provides instruction almost entirely in English, but in a self-contained classroom consisting only of English language learners (ELLs).
- ESL pullout programs supplement regular, mainstream classroom instruction with instruction in a small-group setting outside the mainstream classroom aimed at developing English language skills.
- The sink-or-swim approach provides mainstream classroom instruction with no special help or scaffolding.
- Transitional bilingual education initially delivers instruction and develops students' literacy in the students' native language but puts a priority on developing students' English language skills.
- Two-way bilingual education(also known as two-way immersion) is designed to develop fluency in both the students' first language and a second language; teachers deliver instruction in both languages to classes consisting of both native English speakers and speakers of another language (most commonly Spanish).
- Bilingual maintenance programs generally consist of non-English speakers and, like two-way bilingual education programs, place equal emphasis on maintaining students' primary language and developing their English proficiency.
Notice the order in which I have listed these programs. According to my own research and my reading of others' research, this list proceeds from the most effective to the least effective approaches in terms of helping students become proficient at speaking, writing, and learning in English. This research indicates that in general, the most effective way for students to learn a second language and to learn subject matter in that second language is to learn in the second language—as in the first three programs—rather than learn in the students' native language, as in the last three programs (see Baker & de Kanter, 1981, 1983; Genesee, 1976, 1987; Gersten, Baker, & Otterstedt, 1998; Lambert & Tucker, 1972; Rossell, 2002, 2003, 2004; Rossell & Baker, 1996a, 1996b).
I am aware that this conclusion is highly controversial. In the past, bilingual education has enjoyed enormous support among many researchers and educators. But the apparently successful implementation of sheltered English immersion in California, Arizona, and Massachusetts may change the common perception.
When Is “Bilingual” Not Bilingual?
Despite the common belief in the effectiveness of bilingual education, my observations and my analyses of data from state department of education Web sites indicate that only a minority of immigrant children in the United States are enrolled in bilingual programs in any form. In California, only about 29 percent of English language learners were enrolled in bilingual education in 1998, the year in which this approach was voted out as the default assignment for such students. Approximately 71 percent of California's English language learners participated in programs that used English as the dominant language of instruction—most of them in sink-or-swim or near-sink-or-swim situations (Rossell, 2002). Similarly, in Arizona in 2000 and in Massachusetts in 2002—the years in which these states mandated a switch to structured immersion—only 40 percent of English language learners at most were enrolled in bilingual education (Arizona Department of Education, 2004; Massachusetts Department of Education, personal communications, 2004).
Indeed, despite the lack of intellectual support for the sink-or-swim method, it seems to be the dominant approach to educating English language learners throughout the United States—perhaps because educators believe that the benefits of integration and language role modeling by fluent English speakers outweigh the disadvantages of students' initial noncomprehension of the curriculum, or perhaps because it is simply easier.
Another approach, sheltered English immersion (also called structured immersion), similarly predominates in more schools than one would assume from looking at statistical reports. A sheltered English immersion classroom differs from a mainstream, sink-or-swim classroom because the class is composed entirely of English language learners and is taught by a teacher trained in second-language acquisition techniques. The teacher conducts instruction almost exclusively in English, but at a pace students can keep up with.
Many programs throughout the United States identified as “bilingual education” can be more accurately described as sheltered English immersion because they are actually taught completely or almost completely in English. For example, during the two decades I have spent observing bilingual classrooms across the country, I have observed many Chinese “bilingual education” programs—but have never seen one taught in Chinese. Teachers in these classes believe that Chinese reading and writing skills are not transferable to English because the two written languages are so different. Teachers seldom even teach orally in Chinese because spoken Chinese encompasses many dialects, and it is rare that all students in a classroom speak the same one.
In fact, after observing numerous Russian, Vietnamese, Chinese, Khmer, Haitian, Cape Verdean, Spanish, Japanese, Hebrew, and Portuguese “bilingual education” classrooms and talking with their teachers, I have concluded that schools almost never offer bilingual education that fits the theoretical model, in which students learn to read and write initially as well as learn subject matter in their native language. The sole exception is in languages that use a Roman alphabet. If the primary language doesn't use the Roman alphabet, teachers perceive the transferability of reading skills as too small to justify the effort.
These practical reasons—ignored in the theoretical literature—account for the fact that in the United States, non-Spanish “bilingual education” programs are actually sheltered English immersion programs. This also means that statistics on bilingual education enrollment consistently overestimate the number of students who actually receive native language instruction.
Sheltered English immersion also travels under other labels, such as content ESL and, at the secondary level, sheltered subjects. I once visited a school in New York City that, according to the board of education Web site, had a Bengali bilingual program. When I arrived at the classroom door, however, I found a sign that said Content ESL. In this classroom, Bengali-speaking English language learners were taught by a teacher who was fluent in Bengali. Students who had little English fluency spent most of the day in this class learning English and learning subjects through English. The teacher taught no Bengali at all; he claimed that he did not even use it orally to clarify or explain. These students were actually in a sheltered English immersion class that tailored instruction to their needs.
At the secondary level, many students receive sheltered English immersion in the form of sheltered subject classes (such as sheltered algebra and sheltered U.S. history). Sheltered subject classes have been around for decades, but they often go unnoticed because the language of instruction is English and the curriculum is similar to that of a mainstream classroom. In a sheltered algebra class, for example, the teacher would teach algebra in English to a class composed solely of English language learners.
Although the literature specifies a number of ways in which sheltered English immersion classes differ from mainstream classes (Echevarria & Graves, 2002; Haver, 2002), I have observed many of these classrooms and have seen little difference between the two. Teachers in sheltered English immersion classes seem to speak no more slowly than those in mainstream classes do, and they do not use more visual props. The teachers tell me that the major difference is that they cover less material and use more repetition. Some of these sheltered classes are called “bilingual” if all the students have the same country of origin, but only Spanish speakers in secondary bilingual classes ever hear more than a minimal amount of their native language used in instruction.
Lessons from California
Although sheltered English immersion has been around for decades under various labels, it became the default assignment for English language learners by state mandate in California in 1998, in Arizona in 2000, and in Massachusetts in 2002. Research and observation in California yield some valuable insights about the ways in which teachers implement instruction for their English language learners and the relative effects of the bilingual education and sheltered immersion approaches.
Responding to the Research
In response to the California law (Proposition 227), schools developed two structured immersion models that differ by the ethnic composition of the classrooms and by the amount of sheltering provided. Programs serving English language learners from a variety of linguistic backgrounds provide instruction and conversation in English only. Programs serving exclusively Spanish-speaking students, however, often use Spanish to explain or clarify concepts.
Because the school districts do not reliably distinguish between these different models, evaluating the academic impact of sheltered English immersion is difficult, if not impossible. We can, however, compare with some confidence the academic outcomes of keeping or dismantling transitional bilingual education because the California, Arizona, and Massachusetts laws all allow a school to offer bilingual education to students if the students' parents sign a waiver and if the school can justify using this approach on pedagogical or psychological grounds.
Approximately 10 percent of English language learners in California are still enrolled in bilingual education. My analyses (Rossell, 2002, 2003) show that after controlling for student and school characteristics, the average score increased by six points in reading and by three points in mathematics in schools that eliminated bilingual education. This is a .56 standard deviation gain in reading (a large effect) and a .21 standard deviation gain in math (a small effect). Bali (2001) found that taking Pasadena students out of bilingual education increased their reading scores by two points (.18 standard deviation) and their math scores by one-half point (.03 standard deviation) compared with ELLs who had always been in English immersion classes.
Testing rates are another measure of the effectiveness of alternative programs because a lower testing rate means that the school considers more students unready to take the test. My research (Rossell, 2002) found that schools with more than 240 ELLs enrolled in bilingual education had lower testing rates in reading and math than did those with no ELLs enrolled in bilingual education, after controlling for student and school characteristics. Bali (2000) found that prior to 1998, the rate of testing for English language learners enrolled in bilingual education was 50 percent, compared with 89 percent for those enrolled in English language classrooms. Los Angeles Unified School District found that after five years of participating in the program, only 61 percent of ELLs enrolled in bilingual education were tested, compared with 97 percent of those in English language classrooms (1998).
Unfortunately, there is no scientific research that directly compares the success rates of English language learners in a sheltered English immersion classroom with the success rates of ELLs in a mainstream classroom with ESL pullout. Nevertheless, I believe that at least for the first year, a sheltered classroom is a better environment for most English language learners than a mainstream classroom. My interviews in California indicated that teachers who formerly taught bilingual education but who now teach in sheltered English immersion programs believe the same.
After the first year, however, most English language learners are probably better off in a mainstream classroom with some extra help. Most of them will know English well enough that a sheltered English immersion classroom would slow them down unnecessarily, particularly when new students without any English skills enter the class. The one-year time limit (“not normally intended to exceed one year”) is part of the sheltered English immersion laws in California, Arizona, and Massachusetts and is a provision my fellow researcher and I recommended in our writing (Rossell & Baker, 1996a, 1996b).
Teacher Implementation
My observations of almost 200 classrooms in California from spring 1999 through fall 2004 identified several themes that provide insight into the effectiveness of sheltered English immersion in the state.
Former Spanish bilingual education teachers were impressed by how quickly and eagerly their Spanish-speaking English language learners in kindergarten and 1st grade learned to speak and read in English and how proud the students were of this accomplishment. The teachers were also surprised at how much they themselves liked teaching in sheltered English immersion classrooms, although they had never worked harder (see Haager, Gersten, Baker, & Graves, 2001).
When I asked the teachers in 2001 whether they would ever want to return to teaching in a bilingual education classroom, all of them said no (Rossell, 2002). Bilingual education was a good theory, they claimed, but in practice it had too many problems. They attributed these problems to a lack of materials, teachers, and support.
Interestingly, Chinese bilingual teachers saw Proposition 227 as a non-event. Because they had already been teaching in English, nothing had changed for them except that Proposition 227 justified their practices.
Besides moving most English language learners into sheltered immersion programs, Proposition 227 also changed the way Spanish bilingual education programs operated. The teachers with whom I spoke in the remaining Spanish bilingual education classes in spring 1999 said that they were using more English for instruction than they had in the past. They gave two reasons. First, the Proposition 227 vote expressed California's citizens' preference for a greater emphasis on English, and teachers believed that they should respond to the wishes of the people they served. Second, because the law greatly reduced the demand for bilingual classes, there was no guarantee in any specific school that a bilingual class could be assembled for the next grade in the following year. Accordingly, teachers felt the need to prepare their students for the possibility that they could soon be in an English language classroom. Thus, the task of comparing the effectiveness of bilingual education with that of sheltered immersion is further complicated by the fact that the former is less bilingual than it has been in the past.
Instruction in the Target Language Is Key
Despite the strong support for sheltered English immersion that now exists among educators, policymakers, and the public in California, only about half of all English language learners are actually enrolled in such programs. Most of the other half are in mainstream classrooms, and about 10 percent are still in bilingual education, albeit with more use of English than before.
My classroom observations in California indicate that most educators base decisions about how to teach not just on state mandates but also on their assessment of what their English language learners need, the numbers of English language learners in their classes, and their own philosophy. Most teachers with whom I have talked believe that teaching students in English is more important than ensuring that the students are in a sheltered environment (although the state law requires both).
In general, a mainstream classroom that provides extra help seems to be more practical for many schools, and any academic harm caused by such classrooms is apparently not significant enough to be noticeable to most educators or to offset the relative ease with which schools can form such classrooms. After all, most immigrant children in the United States and throughout the world are in mainstream classrooms, and most of them seem to swim, not sink.
