Learning to read in a second language is a complex, and therefore challenging, developmental process that is influenced by a multiplicity of factors and which has many prerequisites. Children who speak a language other than English upon entering kindergarten or first grade require a specific set of instructional steps in order to develop the necessary pre-reading skills, such as phonological and phonemic awareness for the English language, a reasonable oral language proficiency in English, and the necessary background knowledge in order to access meaning from the written text.
Many different studies attempted to evaluate a variety of instructional approaches with regard to the language of instruction and the timing of initial reading instruction for second-language learners. This issue has remained probably one of the most controversial topics of discussion among educators and politicians alike. The research in this area has failed to provide educators with definite answers on which to base their decisions. Nevertheless, recent work by the Committee on the Prevention of Reading Difficulties in Young Children (Snow, et al., 1998) and Thomas and Collier (1997) provide strong evidence that:
Initial reading instruction in the first language is not detrimental to the child’s acquisition of English. On the contrary, initial instruction in the second language can have negative short term and long term impact on student achievement.
Initial literacy instruction should be conducted in the native language whenever possible.
Formal reading instruction in English should be delayed until a reasonable oral proficiency in English is acquired.
Concurrent formal instruction in the native language of the child should be conducted whenever possible so that the child can access on-grade content information in the subject area, complete the first-language development cycle, and continue in the development of the age appropriate cognitive skills. (There are numerous models of bilingual instruction or instruction in two languages: the child’s dominant language and English. The model proposed by Krashen is presented in Appendix B.)
Educators need to be aware of a variety of risk factors that correlate with reading difficulties for language-minority children, while understanding the importance of individual differences in reference to learning English as a second language or learning to read.
School, Program and Classroom Characteristics That Support the Academic Achievement of English-Language Learners 44
Research suggests that it is problematic to investigate the reading development needs of English-language learners (ELLs) through a limited focus on reading instructional strategies. What is instead needed is a broader focus that encompasses a student's overall educational environment. This is because reading outcomes are determined by complex and multifaceted factors; factors that include student background knowledge, and a student's linguistic and cognitive development (Braunger & Lewis, 1997; Snow, Burns, & Griffin, 1998; Thomas & Collier, 1997). Hence, identifying effective strategies for facilitating the reading development of ELLs must involve an identification of those school, program and classroom characteristics that influence a student's overall academic achievement. Therefore, in this chapter we review what is currently known about both the relative effectiveness of various program models for ELLs (such as late- and early-exit bilingual education), and the school and classroom characteristics that promote the academic achievement of these students.
Section One: Relative Effectiveness of Program Models
Since the 1970s, there has been considerable debate over the relative efficacy of various program models for the education of ELLs, with the varying degree to which ELLs' native languages are used for instruction in these models being the primary source of contention (see Box 3:1 for a list of these models and their descriptions).45 As will be expanded upon in the next section however, this focus on broad programmatic categories may be overly simplistic. Instead, a more sophisticated approach is to include language of instruction as one consideration while evaluating the relative effectiveness of individual programmatic components (August & Hakuta, 1997; Thomas & Collier, 1997). Nonetheless, because language of instruction is a significant factor in the education of ELLs, and because so much attention has been given to the debate over program models, we have chosen in this section to review the research findings regarding these models.
This section begins with a review of three of the four national-level evaluations that have been conducted of programs for ELLs. We then examine key reviews of smaller-scale program evaluations. In both instances, attention is paid to the criticisms that have been made of these evaluations and reviews regarding the validity of their findings. The section concludes with a more detailed discussion of the fourth national-level evaluation: Thomas and Collier (1997). We focus on this study because it appears to have provided significant insight into the relative effectiveness of various program models, while avoiding a number of the limitations of other studies in this area.
PROGRAM MODELS FOR THE EDUCATION OF
ENGLISH-LANGUAGE LEARNERS IN THE U.S.46
Structured immersion: All students in the program are ELLs, and are usually (though not always) from different language backgrounds. Instruction is provided in English, with an attempt made to adjust the level of English so that the subject matter is comprehensible. Typically there is no native-language support.
Early-exit bilingual education: Most students in the program are ELLs. Some degree of instruction is provided through students' native languages; however, because the goal of the model is to transition students into mainstream classrooms as quickly as possible (usually within 2 to 3 years), even within the program there is a rapid shift toward using primarily English. This program type is alternately termed transitional bilingual education.
Late-exit bilingual education: Most students in the program are ELLs who share the same first language. A significant amount of instruction is provided through the students' native language. Like the early-exit model, this model aims to develop English proficiency; unlike the early-exit model however, it has the concomitant goal of developing academic proficiency in the native language. Late-exit bilingual education also differs from the early-exit model in that the former generally involves students for several more years than the latter. This program type is alternately termed maintenance or developmental bilingual education.
Two-way developmental bilingual education: About one half of the students in the program are native English speakers, and the other half are ELLs with the same first language. The goal of this model is to develop proficiency in both languages for both groups of students. Like late-exit bilingual education, this model usually involves students for several more years than the early-exit model.
COMMONLY EMPLOYED ENGLISH-LANGUAGE INSTRUCTIONAL APPROACHES WITHIN THESE PROGRAM MODELS
English as a second language (ESL): Students receive specified periods of instruction aimed at the development of English-language skills, with a primary focus on grammar, vocabulary and communication rather than academic content areas.
Content-based ESL: Students receive specified periods of ESL instruction that is structured around academic content rather than general English-language skills.
Sheltered instruction: Students receive subject matter instruction in English, modified so that it is accessible to them at their levels of English proficiency.
There have been four large-scale, national-level evaluations of programs for ELLs: the American Institutes for Research Study (1978), the Longitudinal Study (1984, 1989), the Immersion Study (1991) and Thomas and Collier (1997). The first three have proven to be of little utility in comparing the effectiveness of program models (August & Hakuta, 1997). However, because they have played such a prominent role in the debate over program models we provide a brief review of them below. We follow this review with a summarization of the findings of a National Research Council report (Meyer & Fienberg, 1992) that evaluates both the Longitudinal and Immersion studies.
The first large-scale, national-level evaluation of programs for ELLs was conducted by the American Institutes for Research, and is commonly referred to as the AIR study (Dannoff, 1978). The study compared students enrolled in Title VII bilingual programs to comparable students not enrolled in such programs. In the study, 8,200 children were measured twice during the school year on English oral comprehension and reading, Spanish oral comprehension and reading, and mathematics. In general, the results showed that students in bilingual education programs failed to gain more on these measures of academic achievement than students not in such programs (with the exception of Spanish reading achievement).
However, the study was the subject of a great deal of criticism. One major criticism involved the questionable strength of the treatment/control group comparison (August & Hakuta, 1997). A second common criticism was the fact that no attempt was made to separate high-quality Title VII programs from those of low quality, and therefore the findings are not informative as to the actual impact of high-quality Title VII programs (Cziko, 1992).
In part because of the ambiguity of the conclusions from the AIR study, two major longitudinal studies were commissioned by the U.S. Department of Education to examine program effectiveness (August & Hakuta, 1997). The first of these studies was the National Longitudinal Evaluation of the Effectiveness of Services for Language Minority Limited English Proficient Students (Burkheimer, Conger, Dunteman, Elliott, & Mowbray, 1989; Development Associates, 1984). Commonly referred to as the Longitudinal Study, it was divided into two phases. The first phase examined the variety of services provided to ELLs (Development Associates, 1984). The second phase attempted to determine the relative effectiveness of programs for these students (Burkheimer, et al., 1989). This latter phase included twenty-five schools, with students from kindergarten to fifth grade being followed over 3 years. The most pertinent of its findings were:
The yearly achievement of ELLs in math and English language arts is facilitated by different approaches, depending on student background factors. For example, students who are relatively English proficient are better able to benefit from English language arts instruction given in English, whereas students who are weak in English or strong in their native language show better yearly English language arts achievement when instructed in their native language.
In earlier grades yearly mathematics achievement gains, as measured by tests conducted in English, can be realized regardless of whether the language used for mathematics instruction is English or the student's native language. In later grades however, yearly mathematics achievement gains will not be realized on tests conducted in English until the child gains some mastery of the language (through English language arts instruction or exposure to instruction in English in other courses, particularly mathematics).
The second study commissioned by the U.S. Department of Education to examine program effectiveness was the Longitudinal Study of Immersion and Dual Language Instructional Programs for Language Minority Children (Ramirez, Yuen, Ramey, & Pasta, 1991). Commonly known as the Immersion Study, it was an attempt at a quasi-experimental longitudinal comparison of three program types: English-only immersion, early-exit bilingual and late-exit bilingual. The study followed over 2,000 elementary school children in nine districts for 4 years. Broadly speaking, the researchers concluded,
Providing LEP [limited English-proficient] students with substantial instruction in their primary language does not interfere with or delay their acquisition of English language skills, but helps them to "catch-up" to their English speaking peers in English language arts, English reading, and math. In contrast, providing LEP students with almost exclusive instruction in English does not accelerate their acquisition of English language arts, reading or math, i.e., they do not appear to be "catching-up." (Ramirez, 1992, p.1)
However, both the Longitudinal and Immersion studies have been widely criticized as having a number of methodological weaknesses that threaten the validity of their findings. The most influential of these criticisms has come from a National Research Council panel of the Committee on National Statistics charged with determining whether the statistical methods employed in the studies were appropriate (Meyer & Fienberg, 1992). The panel's findings included:
The formal designs of both the Longitudinal and Immersion studies were poorly suited to answering the policy questions that appear to have motivated them.
The execution and interpretation of these studies, especially the Longitudinal Study, were hampered by a lack of documentation of the study objectives, the operationalizing of conceptual details, actual procedures followed, and changes in all of the above.
Both studies suffered from excessive reliance on the use of elaborate statistical methods intended to overcome the shortcomings of the research designs.
Although the samples in the Immersion study were followed longitudinally, later comparisons lack validity due to sample attrition.
Reviews of Smaller-Scale Evaluations
As the findings of the National Research Council panel suggest, the manifold limitations of these national-level evaluations have rendered them of little use in comparing the effectiveness of programs for ELLs (August & Hakuta, 1997). In contrast, more has been learned from reviews of smaller-scale evaluations, although these have suffered from methodological limitations as well (August & Hakuta, 1997). Therefore, in the following discussion we examine those reviews of smaller-scale evaluations that have proven most influential in the debate over the effectiveness of program models for ELLs.
The first of these is Baker and de Kanter (1981), which is still "by far the most influential review of studies in bilingual education, despite its age..." (August & Hakuta, 1997, p.144). In this review, the authors attempted to determine whether there was a sufficient research basis to justify the exclusive use of early-exit bilingual education over alternative forms of instruction. To do this they began by locating approximately 150 program evaluations. They then excluded all those they viewed as methodologically unacceptable.47 The findings of the remaining 28 studies were used to compare early-exit bilingual education to other program types in terms of student second-language and math skills outcomes. This was done by: first grouping the studies according to whether they found early-exit bilingual education to have significant positive, significant negative, or non-significant outcomes in relation to comparison programs; and then tallying each of these categories to arrive at an overall summary. Utilizing this technique (commonly termed the voting method), Baker and de Kanter concluded that "the case for the effectiveness of transitional [early-exit] bilingual education is so weak that exclusive reliance on this instruction method is clearly not justified" (p.1). However, it is important to reiterate that these researchers were not attempting to determine whether or not bilingual education was effective, but merely whether there was a sufficient research foundation at that time to justify the exclusive use of early-exit bilingual education over alternative forms of instruction.
Two other influential reviews of smaller-scale evaluations are Rossell and Ross (1986) and Rossell and Baker (1996). Expanding on the work of Baker and de Kanter (1981) and Baker and Pelavin (1984), these researchers considered studies that compared the effects of early-exit bilingual education to those of other educational alternatives for ELLs. Like Baker and de Kanter (1981), these reviews utilized studies they deemed to be methodologically acceptable48 to compare programs in terms of student second-language and math skills outcomes (though Rossell and Baker added the additional consideration of second-language reading outcomes). Also like Baker and de Kanter (1981), these reviews employed the voting method to summarize study findings. Utilizing this method, both Rossell and Ross (1986) and Rossell and Baker (1996) arrived at essentially the same conclusions: (a) The research does not support early-exit bilingual education as a superior form of instruction for ELLs, and (b) structured immersion is a more promising approach.
However, the methodology utilized by both Rossell and Ross (1986) and Rossell and Baker (1996) has been criticized by a number of researchers. One argument is that the selection criteria employed in these reviews for the inclusion of studies are improper (e.g., Greene, 1998; Krashen, 1996; Thomas & Collier, 1997). A second argument is that the voting method utilized in these reviews is error prone (e.g., Greene, 1998; Thomas & Collier, 1997), an assertion to which Rossell and Ross (1986) in fact concede.
A more refined approach than the voting method employed by Baker and de Kanter (1981), Rossell and Ross (1986) and Rossell and Baker (1996) is the use of a technique known as meta-analysis (Thomas & Collier, 1997; Willig, 1985) . A meta-analysis is a statistical synthesis of primary studies that provides a quantitative estimate of the effect of an intervention. Willig (1985) conducted a meta-analysis of the studies reviewed by Baker and de Kanter (1981) and reached a far different conclusion than these authors.49 Utilizing this technique, Willig found that when compared to the provision of no special intervention at all, early-exit bilingual education
consistently produced small to moderate differences favoring bilingual education for tests of reading, language skills, mathematics, and total achievement when the tests were in English; and for reading, language, mathematics, writing, social studies, listening comprehension, and attitudes toward school or self when tests were in other languages. (p.269)50
Yet, although meta-analysis is considered a stronger technique than the voting method, like the other reviews discussed above Willig (1985) has limitations that suggest its findings should be viewed with caution. The most problematic of these limitations is that
it employs the questionable practice of including the same study more than once in the analysis. Willig used a complicated weighing procedure to compensate for this problem, but she may not have been entirely successful in this effort. While the practice of using the same study more than once is quite common in meta-analysis, it does seriously compromise the validity of the inferential statistical analysis. (August & Hakuta, 1997, p.146)
Furthermore, Willig herself notes that although the technique of meta-analysis attempts to statistically control for the methodological inadequacies of the studies analyzed, these inadequacies nonetheless render the results of her analysis "less than definitive"(Willig, 1985, p.269).
However in another, more recent meta-analysis, Greene (1998) provides evidence that supports Willig's findings. Greene conducted a meta-analysis of the studies reviewed by Rossell and Baker (1996) and, like Willig (1985), reached a conclusion that contradicted that of the original review. Greene found that "children with limited English proficiency who are taught using at least some of their native language perform significantly better on standardized tests than similar children who are taught only in English" (Greene, 1998, p.1). Nonetheless, because it is too soon to tell how Greene's analysis will be met by peer review, his findings, like Willig's, should be viewed with caution.51
Thomas and Collier (1997)
Although the use of the meta-analysis technique lends greater credence to the findings of Willig (1985) and Greene (1998) than to those of the other reviews of smaller-scale evaluations discussed above, the evidence they provide in support of bilingual education can not be considered conclusive. Like the national-level evaluations, the reviews of smaller-scale evaluations are plagued by methodological weaknesses and contradictory findings that limit their ability to resolve the debate over program effectiveness. However, the last national-level evaluation we address in this chapter, Thomas and Collier (1997), has avoided a number of the limitations of these studies and reviews. In doing so these researchers have provided strong evidence of the greater effectiveness of two-way and late-exit bilingual education in relation to other program models.
Conducted in five urban and suburban school district sites in various regions of the U.S., Thomas and Collier's research focused on: (a) the length of time needed for ELLs to reach and sustain on-grade-level achievement in their second language; and (b) the student, programmatic and instructional variables that influence language-minority students' academic achievement. The study involved over 700,000 student records collected between 1982 and 1996, including those of over 42,000 students who attended the subject schools for 4 years or more. The data analysis matched the historical records of student background variables and educational program treatment variables with outcome measures, in a series of cross-sectional and longitudinal analyses.
The study's findings in regard to the relative efficacy of various program models are graphically depicted in Figure 3:1. In this figure, the dotted line at the 50th normal curve equivalent (NCE) represents typical native-English speakers' performance on standardized tests in English reading (while making 10 months of progress with each 10-month year of school). Each solid line represents the typical academic performance on these English reading tests of language-minority students participating in the given program type. Although not reflected in Figure 3:1, the same general pattern of language-minority student performance was also manifested on standardized tests in social studies, science, mathematics and English language arts (although achievement in the latter two categories is slightly higher for all program types). As can be seen in Figures 3:1 and 3:2, only the two-way developmental bilingual and late-exit bilingual program models succeeded in producing ELL achievement that reaches parity with that
Figure 3: PATTERNS OF K-12 ENGLISH LEARNERS' LONG-TERM
ACHIEVEMENT IN NCEs ON STANDARDIZED TESTS IN ENGLISH
READING COMPARED ACROSS SIX PROGRAM MODELS52
Program 1: Two-Way Developmental Bilingual Education (BE)
Program 2: Late-Exit BE, including ESL taught through academic content*
Program 3: Early-Exit BE, including ESL taught through academic content*
Program 4: Early-Exit BE, including ESL, both taught traditionally
Program 5: ESL taught through academic content**
Program 6: ESL pullout--taught traditionally
(Results aggregated from a series of 4-8 year longitudinal studies
from well-implemented, mature programs in five school districts)
Copyright 1997, Wayne P. Thomas & Virginia P. Collier
National Clearinghouse for Bilingual Education, The George Washington University
Reproduced by permission of the authors
* We have modified Thomas and Collier's original graph to reflect the terminology of this document. In the original, Late- and Early-Exit BE were termed One-Way Developmental BE and Transitional BE respectively.
** Program 5 corresponds to "structured immersion" as defined in this document.
of native-English speakers. Importantly, the data analysis used to generate Figure 3:1 strongly controlled for student background variables. For instance, the graph only represents cohorts of students who started kindergarten in the U.S. with no proficiency in English and who are of low socioeconomic status (SES) (although the same general patterns were found regardless of student SES). Furthermore, only analyzed were the performances of language-minority students who received just one type of special program, and who only participated in these programs during their elementary school years.53
As was mentioned, Thomas and Collier (1997) avoided a number of the limitations inherent in much of the other research in this area. One example of this that Thomas and Collier only incorporated data from well-implemented examples of the various program models, in order to reduce the confounding effects of implementation differences on instructional effectiveness. Failure to address these confounding effects has been a criticism of the AIR study among others. Another example is the fact that by tracking the progress of over 42,000 students for 3 years or more, Thomas and Collier limited the problems associated with experimental mortality that have undermined the validity of other studies (e.g., Ramirez, et al., 1991).54
However, the most significant difference between Thomas and Collier (1997) and other studies of the effectiveness of program models for ELLs is the former's long-term perspective that addresses students' entire school careers. In contrast to Thomas and Collier's work, other research in this area has generally focused on the early grades, and has rarely tracked student achievement for longer than 4 years. Thomas and Collier argue that this short-term focus on the early grades may be largely responsible for the fact that previous research has often found little difference between program types on the academic achievement of ELLs. This is because, as can be seen in Figure 3:1, the influence of the various program models on student achievement is difficult to detect in the early grades and over short periods of time. However, by addressing student achievement patterns over both the elementary and secondary school years, Thomas and Collier have shown that the differential influence of these models nonetheless becomes increasingly pronounced as language-minority students progress through school (even after they have been exited from special programs).
Thomas and Collier's findings of the superiority of the two-way and late-exit bilingual models in supporting the achievement of language-minority students in reading (and other academic areas) are of serious consequence for Washington State. According to the Office of Superintendent of Public Instruction (1998a), only a small percentage of the ELLs in Washington are served by programs that can be considered late-exit or two-way bilingual education. Instead, almost all the ELLs that are served by special programs are in ones that conform to one of the four models Thomas and Collier found to fail to produce long-term language-minority student achievement that reaches parity with that of native-English speakers.55 These findings suggest that an effective approach to facilitating the reading development of Washington's ELLs would be to provide more of these students with late-exit and two-way developmental bilingual education. Return to Table of Contents
Section Two: Characteristics of Effective Schools and Classrooms
Thomas and Collier's findings should not be underestimated in their significance to the debate over program models for ELLs. However, as was noted previously, the debate itself may be overly simplistic. As August and Hakuta (1997) point out, programs are not unitary but a complex series of components; programs that share the same nominal label can vary greatly, both in terms of these underlying components and in terms of student achievement outcomes. A more sophisticated approach to finding effective methods of educating ELLs is to go beyond a debate over broad programmatic categories, to an effort to identify those school- and classroom-level factors that effect the academic achievement of these students (August & Hakuta, 1997).
Therefore, in this section we provide a synthesis of the findings of studies that attempt to identify school- and classroom-level factors related to effective schooling for language-minority students. We begin with an overview of the types of studies included in this synthesis. We then conclude with a discussion of those attributes of schools and classrooms that have been found to support the academic achievement of language-minority students.56 These are:
positive classroom and school-wide climates;
the use of effective grouping strategies;
the customization of learning environments;
the use of native languages;
the use of instructional strategies that enhance understanding;
the provision of cognitively complex, on-grade-level instruction;
the provision of a balanced curriculum;
the provision of ample opportunities for students to practice English;
The studies included in this review fit into five general categories when classified by the type of methodology employed:
Prospective case study design: In this design, researchers attempt to document changes in schools, programs or classrooms as they occur, and the effects of these changes on student achievement.
Effective schools/classrooms design: In this design, one or more schools or classrooms are identified as effective based on measures of student learning or achievement (i.e., outcome measures). Researchers then attempt to document those aspects of the school(s) or classroom(s) that contribute to this effectiveness.
Nominated schools design: In this design, one or more schools are identified as effective based on the professional judgments of knowledgeable educators, rather than on the basis of outcome measures. As in the effective schools/classrooms design, researchers then attempt to document those aspects of the school(s) that produce this effectiveness.
Experimental design: In this design, treatments (e.g., instructional methods and grouping strategies) are tested on students to determine their effectiveness. This is accomplished by: generating comparison and control groups through randomized assignment, providing the comparison group(s) with the treatment, and then measuring student learning or achievement to determine the treatment's effect.
Quasi-experimental design: Like the experimental design, in this design treatments are evaluated by outcome measures to determine their effect on student learning or achievement. However, this design lacks randomized assignment to treatment and control groups, instead employing methods to approximate the statistical strength of randomized assignment.
Each of these design types has its methodological strengths and limitations. Because a study's limitations do not necessarily invalidate its findings, we have included all of these design types in this section.58 However, the reader should remain cognizant of the limitations of the nominated schools design in particular. Although the nominated schools studies provide highly detailed descriptions of effective schools, their general lack of student achievement measurements greatly undermines their authors' ability to make assertions as to what attributes make these schools effective, or even that they are truly effective at all (August & Hakuta, 1997).
Table 3:1 categorizes the central studies reviewed in this section according to their design type. Aside from the studies listed therein, we also incorporate other research when doing so helps to inform our discussion.
A number of researchers report that positive classroom (Pease-Alvarez, et al., 1991; Tharp, 1982) and school-wide (Berman, et al., 1995; Carter & Chatfield, 1986; Goldenberg & Gallimore, 1991; Thomas & Collier, 1997) climates are correlated with higher academic achievement among language-minority students. Although the particular manifestations of such climates vary, several commonalties are apparent. Throughout the research, schools and classrooms with positive climates are depicted as those in which:
Significant value is placed on the linguistic and cultural backgrounds of language-minority students (Berman, et al., 1995; Lucas & Katz ,1994; Pease-Alvarez, et al., 1991; Thomas & Collier, 1997). This validation of students' linguistic and cultural backgrounds is believed to enhance student self-esteem, thereby supporting student achievement.
There are high expectations in regard to the academic success of language-minority students (Berman, et al., 1995; Carter & Chatfield, 1986; Goldenberg & Gallimore, 1991; Moll, 1988). Rejected is the idea that language-minority students are intellectually or academically disadvantaged because of their socioeconomic status, or because of their cultural or linguistic backgrounds. Indeed, a student's home language and culture are viewed as resources to be built upon as opposed to liabilities to be remediated. Students are seen as being able to achieve as much as instruction and the curriculum allow. Therefore, poor student academic performance is considered a failure on the part of the school, not of the home environment.
Students experience a safe environment that is conducive to learning (Carter & Chatfield, 1986; Garcia, 1988; Pease-Alvarez, et al., 1991; Thomas & Collier, 1997). On a school-wide level this environment is characterized as safe, orderly and courteous (Carter & Chatfield, 1986). On a classroom level, it is characterized as one in which teachers foster a friendly, almost familial classroom atmosphere that encourages trusting and caring relationships between class members (particularly between students and their teacher) (Garcia, 1988; Pease-Alvarez, et al., 1991).
In addition, schools depicted as having positive climates are characterized as those in which language-minority students, and the programs implemented to serve their needs, are an integral part of the overall school operation (Carter & Chatfield, 1986; Pease-Alvarez, et al., 1991).60
Effective Grouping Strategies
A recurrent theme in the studies reviewed in this section is the effectiveness of utilizing student grouping strategies such as cooperative learning groups, peer tutoring and partner reading in the classroom (Berman, et al., 1995; Dianda & Flaherty, 1995; Garcia, 1988; Henderson & Landesman, 1992; Lucas & Katz, 1994; Moll, 1988; Pease-Alvarez, et al., 1991; Short, 1994; Slavin & Madden, 1995; Slavin & Yampolsky, 1992; Tharp, 1982; Thomas & Collier, 1997; Wong Fillmore, et al., 1985). The most routinely cited of these techniques is cooperative learning. Although there are many variations to the cooperative learning model, the elements most frequently cited as distinguishing it from traditional, whole class instruction are:
heterogeneous groups of two to six students;
lessons structured in such a manner that students depend on each other in a positive way for their learning;
an explicit focus on interpersonal and small group skills; and
teachers as consultants or facilitators of learning, as opposed to transmitters of the material.
In a review of studies done on cooperative learning, Slavin (1995) notes that it "is one of the most extensively evaluated of all instructional innovations" (p.19). His analysis of ninety experimental and quasi-experimental studies concludes that cooperative learning has a positive effect on student achievement and race relations. Furthermore, the studies indicate the overall effects on "student self-esteem, peer support for achievement, internal locus of control, time on-task, liking of class and classmates, cooperativeness, and other variables are positive and robust" (Slavin, 1995, p.70). Other researchers have also suggested that, beyond the benefits listed by Slavin (1995), cooperative learning techniques can be effective in supporting second-language acquisition (especially in classrooms that incorporate native speakers) (Bejarano, 1987; Cohen, DeAvila, & Intiti, 1981, as cited in Kagan, 1986; Jacob, Rottenberg, Patrick, & Wheeler, 1996; Sharan, Bejarano, Kussell, & Peleg, 1984).
Customized Learning Environment
Studies have found that in schools and classrooms that effectively support the academic achievement of language-minority students, educational approaches are customized to meet the diverse needs of their students (Berman, et al., 1995; Moll, 1988; Short, 1994; Tharp, 1982; Tikunoff, 1983; Wong Fillmore, et al., 1985). Understood is that language-minority students are not a homogenous group and that there is no educational panacea for them. For instance, Berman, et al. (1995) note that in the effective schools they studied,
All the language development programs were flexibly constructed to accommodate students with varying levels of fluency and, where appropriate, students from different language backgrounds. Rather than trying to fit all the LEP students into one mold, teachers could adjust curriculum, instruction, and the use of primary language to meet the varying needs of students. Such flexibility is necessary because of the diversity of students... (Berman, et al., 1995, no page number)
Similarly, Moll (1988) observes that although the effective teachers he studied held similar views about teaching, they nevertheless developed distinct instructional programs to accommodate the particular needs of their students.
Tikunoff, (1983), Tharp (1982) and Wong Fillmore, et al. (1985) conclude that this customization of educational approaches should also include efforts to make instruction compatible with a student's culture. This argument is consistent with the work of a number of researchers who assert that: discontinuities between a student's home culture and that of his or her school can result in academic difficulties; and, conversely, instructional approaches that are more consistent with a particular ethnic group's norms of social and linguistic interaction help support academic achievement through diminishing these discontinuities (e.g., Guild, 1998; Pease-Alvarez & Vasquez, 1994; Zanger, 1991). A commonly cited example of such an instructional approach is cooperative learning, which is believed to be more consistent with the social norms and learning styles of certain ethnic groups than is traditional, whole-class instruction (e.g., Tharp, 1982).61, 62
Use of Native Languages
Several studies reviewed in this section note the advantage of incorporating a student's home language into his or her educational experiences (Berman, et al., 1995; Lucas & Katz, 1994; Pease-Alvarez, et al., 1991; Thomas & Collier, 1997; Tikunoff, 1983). For instance, Thomas and Collier (1997) found that of all the variables analyzed in their major study of programs for ELLs, first-language support "explains the most variance in student achievement and is the most powerful influence on [language-minority] students' long term academic success" (p.64). Even Lucas and Katz (1994), in a study of nine exemplary programs that were ostensibly English-only, nevertheless found these programs to incorporate the use of native languages in a variety of ways. Such incorporation is believed not only to support student self-esteem (as discussed above), but to make academic content more accessible, allow for more effective interaction between students and their teachers, and provide greater access to students' prior knowledge (Lucas & Katz, 1994). Furthermore, the use of a student's home language is also believed to indirectly facilitate the development of English-language proficiency (see Chapter One).
Instructional Strategies that Enhance Understanding
A number of studies have found that effective teachers of ELLs utilize specially tailored strategies to enhance student understanding (Berman, et al., 1995; Chamot, et al., 1992; Dianda & Flaherty, 1995; Henderson & Landesman, 1992; Moll, 1988; Muniz-Swicegood, 1994; Rosebery, et al., 1992; Short, 1994; Tharp, 1982; Thomas & Collier, 1997; Wong Fillmore, et al., 1985). These strategies include: a focus on hands-on, experiential learning; teaching students meta-cognitive strategies; and connecting the curriculum to a student's culture and experiences. Also, in addition to these generally applicable strategies, a number of strategies are cited that focus primarily on facilitating understanding when instruction is in English. These include: the use of manipulatives, pictures, objects and films related to the subject matter; providing demonstrations of academic tasks before students are asked to perform them; having a predictable daily classroom schedule; and the modification of teacher speech in a manner appropriate to a student's current level of English proficiency.63
Cognitively Complex, On-Grade-Level Instruction
Research has shown that the schooling of students from working-class families (a group that is inclusive of most language-minority students) is disproportionately focused on intellectually limited, low-level skills (e.g., Anyon, 1980, 1981; Goldenberg, 1984, 1990, as cited in Moll, 1988; Oakes, 1986). Furthermore, the work of Diaz, Moll, and Mehan (1986) and Moll (1986) has found that there is the additional tendency to reduce the curriculum's level of complexity even further in order to match ELLs' level of English proficiency. However, many of the studies reviewed in this section have found that supporting the academic achievement of language-minority students requires a reversal of these conditions (Berman, et al., 1995; Goldenberg & Gallimore, 1991; Moll, 1988; Thomas & Collier, 1997; Wong Fillmore, et al., 1985). The most significant of these is Thomas and Collier (1997), which found that for a program to succeed in ELLs reaching long-term academic achievement levels that are equal to the average achievement of native-English speakers, it must provide:
cognitively complex, on-grade-level instruction through students' first language for as long as possible (at least through grade 5 or 6); and
cognitively complex, on-grade-level instruction through English for part of the school day, utilizing content-based ESL or sheltered instruction methods.64
It is important to note Thomas and Collier's assertion that when instruction is provided in English, content-based ESL and sheltered instruction should be employed. Content-based ESL is ESL instruction that is structured around academic content rather than general English-language skills. Sheltered instruction is subject matter instruction in English that is presented in a manner that makes it accessible to ELLs given their levels of English proficiency (e.g., teacher speech is modified and greater emphasis is placed on hands-on activities). Both methods are widely believed to facilitate the provision of cognitively complex, on-grade-level instruction to ELLs (e.g., Krashen, 1998; Thomas & Collier, 1997).65
Of particular importance is the utilization of these methods when a school can not provide an ELL with content instruction through his or her first language. Too often in these situations, ELLs are placed in mainstream classrooms for most of the day and only provided with special instruction in the form of traditionally taught ESL pull-out sessions. Because of the linguistic barriers ELLs face in such programs, they are effectively denied full access to content instruction until high levels of English-language proficiency are achieved. This is particularly problematic because, as Cummins (1994) states,
Language learning is a process that takes time; ESL students may require five (or more) years to catch up with their native English-speaking peers in academic aspects of English. Clearly, ESL students' cognitive growth and their learning of subject matter content cannot be postponed until their English-language skills are developed to the level of their classmates'. (p.56)
However, by modifying English-only programs so that content-based ESL and sheltered instruction are provided, the cognitive growth and content learning of ELLs can be supported while these students acquire a level of English proficiency that allows them to participate fully in mainstream classrooms (Krashen, 1998).
Several studies report the benefits of classroom instruction that incorporates both a skills-based approach to pedagogy and a more holistic, meaning-based approach (Dianda & Flaherty, 1995; Goldenberg & Gallimore, 1991; Pease-Alvarez, et al., 1991; Slavin & Madden, 1995; Slavin & Yampolsky, 1992; Tharp, 1982).66 The skills based approach is characterized by the explicit teaching of discrete skills such as phonics, word recognition and specific comprehension skills. Researchers cite this approach as being beneficial for highly structured skill and knowledge domains such as mathematical computation, map reading and explicit reading comprehension strategies (Goldenberg, 1991). The holistic approach, on the other hand, focuses on student engagement with meaningful material that is being taught for comprehension. This approach supports the acquisition of higher order conceptual and linguistic skills, and is best suited for less structured skill and knowledge domains such as reading comprehension (Goldenberg, 1991).
Methods discussed in the literature for incorporating the holistic, meaning based approach to instruction in the classroom include:
Word wall: This method involves regularly posting words on a classroom wall, which can then be incorporated into various activities to support students' literacy development. The words can be selected from students' writing, books that students are reading, or other sources related to students' learning.
Language experience approach: In this method, students dictate their own stories or experiences to the teacher. This text is then used as reading material for its author, and is also sometimes used as reading material for other students in the class.
Instructional conversations: These are discussion-based lessons in which the teacher draws from students' prior knowledge, experiences and home culture in order to guide extended, theme-focused conversations that promote analysis, reflection and critical thinking.
Dialogue journals: In this method, the teacher and student engage in an extended written conversation through regular journal entries. Students write as much as they choose on predominately self-selected topics. The teacher responds by making comments, offering observations and opinions, requesting and giving clarifications, asking questions, answering student questions, and introducing new topics. Types of dialogue journals include daily personal journals, literature response journals, content response journals and learning logs.
Writer's workshop: This method teaches writing through student composition of authentic texts. Workshops occur in regular and predictable blocks of time, and include mini-lessons on the mechanics and craft of writing, as well as student generation of texts in a multi-step process. This process includes drafting, conferencing and sharing, revising, redrafting, editing, and the "publishing" of the completed text.
Thematic units: This method involves integrating the curriculum across a variety of subject areas. It often includes teachers deciding on the theme and skills to be developed, yet allowing students to select interesting topics to pursue while studying the broader theme.
Opportunities for Practice
ELLs' exposure to meaning-focused use of the English language, especially during interactions with native-speakers, is believed to be critical to the language acquisition process. It is not surprising therefore that several of the studies reviewed in this section explicitly note that effective schools and classrooms provide ELLs with ample opportunities to engage in communicative interaction using English (Berman, et al., 1995; Garcia, 1988; Moll, 1988; Rosebery, et al., 1992; Wong Fillmore, et al., 1985). The provision of such opportunities is also implicit in many of the other studies, with the utilization by teachers of techniques such as cooperative learning, peer tutoring, instructional conversations and dialogue journals being cited numerous times. These techniques are generally believed to provide significant exposure to meaning-focused use of English, and therefore to facilitate the language acquisition process.67
Collaboration Between Home and School
A prominent theme among the studies reviewed in this section is the importance of school efforts to support student achievement through collaboration with students' families. For instance, a number of studies cite efforts on the part of schools and teachers to encourage parental involvement in their children's education (e.g., helping with homework, conferring with teachers, and participating in school governance) (Berman, et al., 1995; Carter & Chatfield, 1986; Dianda & Flaherty, 1995; Garcia, 1988; Goldenberg & Gallimore, 1991; Pease-Alvarez, et al., 1991; Slavin & Madden, 1995; Slavin & Yampolsky, 1992). The benefits of such parental involvement have been extensively documented throughout the broader educational literature, and include improved academic achievement, enhanced English-language skills, improved in-school behavior, increased cognitive growth and improved home-school relations among others (for reviews of the research on this subject see, for example, August & Hakuta, 1997, and Bermudez & Marquez, 1996).68
Several studies also found that effective schools supported language-minority student achievement through helping the families of these students access needed social services (Berman, et al., 1995; Carter & Chatfield, 1986; Dianda & Flaherty, 1995; Slavin & Madden, 1995; Slavin & Yampolsky, 1992). For instance, Slavin and Madden (1995) note that in the Success for All program they studied, family support staff provided assistance when there were indications that students were not working up to their full potential because of problems at home (such as not receiving adequate sleep or nutrition). In such cases, links with appropriate community service agencies were made to provide as much support as possible for parents and children. Similarly, Berman, et al. (1995) found that the exemplary schools they studied delivered a range of integrated health and social services that reflected a vision of the school as an integral part of the surrounding community.
Effective Staff Development
Staff development is routinely cited as an important component of effective schools for language-minority students (Berman, et al., 1995; Carter & Chatfield, 1986; Chamot, et al., 1992; Garcia, 1988; Goldenberg & Sullivan, 1994; Moll, 1988; Pease-Alvarez, et al., 1991; Slavin & Madden, 1995; Slavin & Yampolsky, 1992; Thomas & Collier, 1997). Unfortunately, these studies fail to provide substantial evidence as to what forms of staff development are most effective. However, other researchers who address this issue argue that teachers who work with language-minority students should receive professional development that includes a focus on:
building respect for students' cultural and linguistic backgrounds, as well as how to accommodate and build upon these in the school and classroom (Burkart & Sheppard, 1995; Collier, 1995; Uranga, 1995);
current theories on how children acquire a second language, and how learning content is affected when mediated through a second language (Burkart & Sheppard, 1995);
how to use multiple, alternative and authentic means of assessing student knowledge and learning (Collier, 1995; Uranga, 1995);69
methods for eliciting student use of the target language, especially cooperative learning (Collier, 1995; Uranga, 1995; Leighton, Hightower, & Wrigley, 1995);
cultivating teachers' disposition toward, and competence in, critically reflecting on their own teaching (Leighton, et al., 1995);
methods for making instruction comprehensible to students, such as connecting the curriculum to students' experiences and cultures, modifying teacher speech, and utilizing hands-on, experiential learning (Burkart & Sheppard, 1995; Collier, 1995; Leighton, et al., 1995); and
methods for infusing language development into content instruction (Leighton, et al., 1995).
Furthermore, a number of researchers have stressed the importance of providing professional development focused on the needs of ELLs to not just bilingual or ESL staff, but to all mainstream teachers with ELLs in their classrooms (e.g., Burkart & Sheppard, 1995; Galbraith & Anstrom, 1995; Leighton, et al., 1995). This is especially important because, as Burkart & Sheppard (1995) note, "The majority of practicing teachers have not been trained to deal with the linguistic and cultural diversity of today's classrooms" (no page number).
Return to Table of Contents
As was discussed in the beginning of this chapter, efforts to support the reading development of ELLs should include a focus on these students' overall academic achievement. This is because reading development is influenced by factors such as student background knowledge, and a student's linguistic and cognitive development (Braunger & Lewis, 1997; Snow, et al., 1998; Thomas & Collier, 1997). Because of this fact, we have reviewed what is currently known about the relative effectiveness of various program models for the education of ELLs, as well as the school and classroom characteristics that promote the academic achievement of these students.
Research suggests that late-exit and two-way developmental bilingual education are superior to other types of special programs for ELLs (e.g., structured immersion) in supporting the academic achievement of these students. The findings of Thomas and Collier (1997) show that these program models succeed in producing average language-minority student achievement in English reading, English language arts and other academic areas that reaches parity with that of native-English speakers, while program models such as early-exit bilingual education and structured immersion do not. However, as was noted above, this focus on broad program categories is overly simplistic. Researchers have pointed out that programs are not unitary but a complex series of components, and that programs that share the same nominal label vary in terms of these underlying components, and in terms of student achievement outcomes. Because of this we have examined those studies that attempt to document school- and classroom-level factors that support the academic achievement of language-minority students. The studies suggest that these factors include the provision of a balanced curriculum, the use of effective grouping strategies, etc. Interestingly, the superiority of two-way developmental and late-exit bilingual education programs may be largely due to the fact that they usually incorporate a number of these school- and classroom level factors more fully than do programs that conform to other models. For instance, two-way and late-exit programs generally utilize native languages more extensively and provide more cognitively complex, on-grade-level instruction.