English language learners—students who come from homes in which a language other than English is spoken—are one of the fastest-growing student populations in U.S. schools. At the same time, the debate over the best way to help English language learners succeed in school has become increasingly heated and political: Some states have abolished or sharply limited native language instruction, whereas others continue it. Today, the stakes are higher than ever. Because many federal and state policies now mandate that schools demonstrate adequate yearly progress of every student subgroup, schools with large English language learner populations face serious consequences unless these students succeed.
The controversy over optimal education for English language learners has focused on beginning reading instruction in particular. Should schools teach English language learners to read in their native language first and then in English, or should English language learners be taught to read in English from the outset with appropriate supports? Whatever the language of instruction, which instructional strategies are most effective? To answer these questions, we must examine the research on beginning reading instruction for English language learners.
Language of Instruction
Teaching reading to English language learners presents a dilemma: How can these students learn to speak and read an unfamiliar language simultaneously? Native English speakers learn to read primarily by applying phonics skills to put sounds together until they match words in their speaking vocabulary (National Reading Panel, 2000). Students put the sounds /c/ /a/ /t/ together to form the word “cat,” for example. But what if they don't know the word “cat”?
Educators agree that the goal of reading instruction for English language learners is success in reading English; nevertheless, they believe in radically different paths to that goal. Advocates of native language, or bilingual, instruction argue that schools must teach reading in the student's native language first; only after the student is proficient in that language and has developed substantial proficiency in spoken English should he or she be transitioned to English-only reading instruction. A related school of thought advocates paired bilingual instruction: English language learners are taught to read in their native language and in English at different times of the day. This idea is at the core of two-way bilingual instruction, in which English language learners and native English speakers both learn to read in both languages (see Calderón & Minaya-Rowe, 2003; Howard, Sugarman, & Christian, 2003).
By contrast, opponents of bilingual education argue that delaying reading instruction in English is counterproductive and that English-only instruction—with such supports as instruction in vocabulary—is a more effective approach. In particular, English-only proponents advocate “structured immersion,” which presents English reading in a systematic, structured sequence (see Gersten, 1985; Rossell & Baker, 1996).
Previous reviews of research on the effects of bilingual education on the reading performance of English language learners have yielded conflicting conclusions. Willig (1985), Wong-Fillmore and Valadez (1986), and Greene (1997) concluded that research favored bilingual programs. Other reviewers (for example, August & Hakuta, 1997) have also given the nod to bilingual education—but have maintained that quality of instruction is probably more important, regardless of which language is taught first (see also Christian & Genessee, 2001; Goldenberg, 1996; Secada et al., 1998). By contrast, Rossell and Baker (1996) reviewed the same literature and concluded that most studies found no difference between bilingual and English-only strategies. The Rossell and Baker review has significantly influenced state policies, resulting in changes opposing bilingual education.
Recently, we carried out a thorough review of the research in an attempt to resolve the conflicting claims (Slavin & Cheung, 2003). We applied consistent standards to the studies: Studies had to compare bilingual instruction to English-only instruction with English language learners; evidence had to show that the two groups were comparable in reading performance before the treatments began; treatments had to have been in place for at least one year; and researchers had to use a quantitative, objective measure of reading performance.
Seventeen studies of elementary reading instruction met these commonsense criteria. The results were surprisingly favorable to bilingual approaches. Most studies found significant positive effects of bilingual education on reading performance, and other studies found no differences—but in no case did positive results from an English-only strategy exceed those from a bilingual strategy. When we could not compute effect sizes, we summarized study outcomes in terms of relative performance of bilingual and English-only programs.
Even more surprising to us was the strength of the evidence in favor of teaching students to read in both their native language and in English at different times of the day. This paired bilingual strategy was common in the 1970s, when many of the studies of bilingual education were conducted, but it survives today primarily in two-way bilingual programs, in which teachers instruct (typically) Spanish- and English-dominant students to read both Spanish and English, from kindergarten or 1st grade onward.
We found three studies of beginning reading in which Spanish-dominant students were assigned at random (for example, by coin toss) to be taught either in English only or in Spanish and English. Plante (1976) randomly assigned Spanish-dominant students in New Haven, Connecticut, to paired bilingual or English-only instruction beginning in 1st grade. The students taught in both languages performed much better on English reading tests than did the students taught only in English. Similarly, Huzar (1973) randomly assigned Spanish-dominant students in Perth Amboy, New Jersey, to either paired bilingual or English-only instruction. Again, students taught in two languages performed much better on English reading measures than did students taught only in English. Maldonado (1994) studied Spanish-dominant students in special education classes in Houston, Texas. Half were randomly assigned to a class in which instruction was primarily in Spanish in kindergarten, in both Spanish and English in 1st grade, and primarily in English in 2nd grade. Other students were randomly assigned to be taught in English only. Results of the English Comprehensive Tests of Basic Skills strongly favored the students taught in both languages.
These three studies are particularly important because in each case students were assigned at random to the two treatments. We may therefore consider the groups of students equivalent in every way except for their being taught in Spanish and English or in English only. Each study involved a small group of students, however; as a result, some of the comparisons are not statistically significant. Nevertheless, researchers consistently found that paired bilingual strategies aided English reading. Perhaps the Spanish reading instruction served as a bridge from the students' strong oral skills in Spanish to English reading. A Spanish-speaking student who learns the alphabetic principle and knows that the sounds /g/, /a/, /t/, and /o/ combine to form the word “gato” (Spanish for “cat”) might be better prepared to read English than a student who is puzzling over what word is formed out of the sounds /c/, /a/, and /t/. Of course, in each study, students in the paired bilingual classes received more time in reading instruction in both English and Spanish. Still, these studies challenge both transitional bilingual education, which delays English reading instruction for several years, and English-only or English immersion instruction.
Other multiyear studies that compared bilingual program students with similar students in English-only programs also generally favored the bilingual programs—although not in every case. Among nine multiyear matched studies, five strongly favored bilingual approaches and four found no differences. Not one favored English-only instruction. Among five one-year studies, all five favored bilingual methods—although in some cases, the differences were quite small. Although several studies offered unclear descriptions of the treatments, most of the studies examined bilingual treatments that must have introduced English reading in 1st grade or earlier because students were performing well in English by the end of 1st grade.
Existing research that met our inclusion criteria indicated that bilingual programs do not harm and in fact usually improve the English reading performance of English language learners. In addition, the research found that paired bilingual programs, in which students learn to read in both languages, seem to hold particular promise. Educators need more research on this topic—particularly large-scale studies in which students are randomly assigned to bilingual and English-only programs.
Effective Reading Programs
In recent years, many research reviewers (for example, August & Hakuta, 1997; Christian & Genessee, 2001) have concluded that methods of instruction may be more important than language of instruction for English language learners. In addition to reviewing research on language of instruction, we also examined research on replicable programs that educators have used to improve the reading achievement of English language learners.
We used the same inclusion criteria for effective program studies that we did for studies of language of instruction, except that the study duration could be as brief as 12 weeks. In this review, we reported outcomes in languages other than English (in practice, this always meant Spanish), as well as English outcomes.
The largest number of experimental studies evaluated the Success for All program. Success for All is a comprehensive reading program that emphasizes systematic phonics, cooperative learning, tutoring for struggling students, family support programs, and many other elements (see Slavin & Madden, 2001). A number of researchers have extensively evaluated the use of both the English and the Spanish bilingual versions of Success for All with English language learners; they have consistently found both versions to be effective in improving English and Spanish reading performance (Hurley, Chamberlain, Slavin, & Madden, 2001; Livingston & Flaherty, 1997; Nunnery et al., 1997; Ross, Smith, & Nunnery, 1998; Slavin & Madden, 1995). Two studies evaluated the use of Direct Instruction—another systematic phonics program taught in English—with English language learners (Becker & Gersten, 1982; Gersten, 1985); both found that Direct Instruction improved students' English reading achievement. A study in England involving students from Bangladesh found that a third phonetic program, Jolly Phonics, was more effective than a comparison method (Stuart, 1999).
Researchers have also found two tutoring programs for struggling beginning readers to be successful with English language learners: Escamilla (1994) evaluated a Spanish adaptation of Reading Recovery; Gunn, Biglan, Smolkowski, and Ary (2000) evaluated a version of Direct Instruction used in small-group and one-to-one tutorials. Finally, Goldenberg (1990) found positive effects of a Spanish home and school literature program on kindergartners' early literacy development in Spanish.
It is interesting to note that all but one of the programs we found to be successful with English language learners were adaptations of programs also found to be successful with English-dominant students. In particular, Success for All (Slavin & Madden, 2000, 2001), Direct Instruction (Adams & Engelmann, 1996), and Reading Recovery (Pinnell, Lyons, DeFord, Bryk, & Seltzer, 1994) have all been extensively evaluated with a wide variety of students. The finding that effective early literacy programs for English language learners are closely related to those for English-dominant students does not mean that English language learners do not need appropriate accommodations to their needs, such as vocabulary instruction, language development, and programs emphasizing cooperative problem solving in small groups (see, for example, Calderón, Hertz-Lazarowitz, & Slavin, 1998; Carlo et al., in press). Most of the adaptations to existing programs include such elements. The data do suggest, however, that one-to-one tutoring and methods emphasizing systematic phonics are as effective for English language learners as they are for other students.
From Research to Practice
The research on reading instruction for English language learners provides several useful guidelines for policy and practice. First, the literature supports the use of native language in early reading instruction, especially for paired bilingual strategies that teach reading both in the native language and in English from 1st grade onward. Although not every study found benefits from bilingual instruction, most studies supported this strategy and the remainder found no differences. None found English-only strategies to be superior. Second, research indicates that paired bilingual strategies—part of most two-way bilingual programs—tend to be effective; this would seem to support the expansion of two-way programs. Educators need more research, however, on two-way programs themselves.
Third, research on beginning reading programs for English language learners finds consistent positive effects of programs that use systematic phonics—in particular, the Success for All and Direct Instruction programs. One-to-one and small-group tutoring models also demonstrated positive effects. Finally, in a broader sense, the research on effective reading programs for English language learners gives educators reason for optimism. These students can learn to read, and replicable programs and policies greatly accelerate their reading progress.
