After Exit : Academic Achievement Patterns of Former English Language Learners

With few exceptions, accountability systems for programs for English language learners (ELLs) have focused on the achievement patterns of ELLs who are still considered “limited English proficient” and program evaluations have been unable to answer the question whether ELLs actually catch up with English proficient peers after attending a bilingual or English as a Second Language (ESL) program. Disaggregating data for former ELLs can therefore provide important information for long-term district and program accountability. The study was concerned with the achievement patterns in English language arts, Math, and Science of former ELLs who attended a bilingual and a English as a Second Language (ESL) program. It also explored whether length of program participation and grade level exited played a significant role in predicting academic achievement patterns for these exited students. Results indicate that 4 grade students more closely paralleled nonELL students’ achievement patterns than 8 grade students, particularly for the BE students. While length of program participation is not a significant predictor of former ELLs’ academic success, exit grade does emerge as an important variable to take into consideration in setting exit guidelines. Introduction The inclusion of English language learners (ELLs) in recent accountability and school reform efforts (e.g., No Child Left Behind) has placed bilingual and English as a Second Language (ESL) programs under renewed scrutiny as to their ability to overcome the linguistic barriers ELLs encounter when entering U.S. schools. With few exceptions, the effectiveness of bilingual/ESL programs has been evaluated on the performance of ELLs still being served through the program. After exit: Academic achievement patterns of former English language learners 2 2 While providing some information about the performance of limited English proficient students, these program evaluations are unable to describe what happens to ELLs once the latter have been placed in a mainstream classroom on a fulltime basis after having been reclassified as “fluent English proficient.” This article is concerned with the question whether ELLs in fact “catch up” academically after they exited from a bilingual or ESL program. It is argued that including former ELLs in accountability planning will provide a better understanding of whether a school or district has been able to provide equal educational opportunities for this group of students. After a brief overview of the schooling of ELLs in the United States, the article discusses exit policies, the role of exit rates in ELL program evaluations, and academic achievement patterns for former ELLs. Then, the results of a study examining former ELLs’ 4 and 8 grade achievement patterns in English language arts, mathematics, and science will be presented. The article concludes with the study’s implications for future program evaluations and policies. Issues of Exit in ELL Programs Background Title VI of Civil Rights Act (1964) requires school districts to take affirmative steps to avoid discrimination against ELLs based on their limited proficiency in English. In the landmark court case Lau v. Nichols (1974) the Supreme Court proclaimed, “there is no equality of treatment merely by providing students with the same facilities, textbooks, teachers, and curriculum; for students who do not understand English are effectively foreclosed from any meaningful education” (cited in Teitelbaum and Hiller, 1977, p. 7). The court refrained, however, from recommending one particular program to remedy such discrimination, pointing out that the intervention could be instruction in the students’ native language or in English. In Castañeda v. Pickard (1981) the court identified three criteria to assess when a district’s intervention can be considered adequate: it must have theoretical merit, it must be implemented with sufficient resources, and it must demonstrate its effectiveness through program evaluation (Crawford, 1999). Not surprisingly then, different programs have been developed and implemented to meet the needs of ELLs over the last three decades. Generally speaking, these programs differ according to the language of instruction (bilingual or English-only), program goals (bilingualism or English proficiency), and the program’s target group (ELLs only or ELLs and native English speakers) (Brisk 1998; Ovando, Collier, and Coombs, 2003). Much of the controversy regarding the schooling of ELLs has focused on whether schools should adopt bilingual or English-only approaches (for detailed overviews of the effectiveness debate, see August and Hakuta, 1997; Collier, 1992; Crawford, 1997; Cummins, 1996; Rossell and Baker, 1996). One implied feature shared by the majority of the programs is their temporary nature: the ultimate goal is to prepare ELLs for placement in a mainstream classroom within a certain time frame. As a result, deciding when and how to exit ELLs from a bilingual or ESL program plays an integral role in the schooling of ELLs. Furthermore, the temporary nature draws attention to the length of time the students remains in a program. Both issues are discussed in more detail below. Exit Policies Little agreement exists as to how to determine eligibility of minority language speakers for specialized language services (e.g., Rossell and Baker, 1988) and, subsequently, on how to decide that 1 Exceptions are maintenance bilingual programs. Perceived and implemented as mainstream programs, they typically do not include an exiting process. Education Policy Analysis Archives Vol. 12 No. 50 3 3 ELLs no longer need bilingual or ESL services. There is substantial variation within and across states in exit policies and guidelines. States and districts differ widely in whether to use a cut-off score on a test and in what cut-off score to select to determine exit-readiness (De Avila, 1990). Further, some states have no guidelines regarding which assessments to use, others either recommend or mandate the use of certain instruments, including standardized achievement tests, English oral language proficiency tests, English literacy tests, or teacher judgment (De George, 1988; Gandara and Merino, 1993, Fleischman and Hopstock, 1993; Rossell and Baker, 1988). Comparison of exited students across states and even across districts is therefore extremely difficult (Linquanti, 2001). The choice of assessment instrument, the use of cut-off scores, and the data used to make placement decisions for ELLs reflect a district’s definition of the linguistic needs of ELLs, the purpose of the ELL program, and expectations for mainstream classroom teachers to be able to work with ELLs (DeAvila, 1990; Nadeau and Miramontes, 1988). For instance, defining ELLs’ needed language proficiency as survival oral language proficiency may translate into an exclusive reliance on oral assessments for exiting and a focus in the ESL program on oral language development rather than literacy development. Similarly, exit policies that use similar criteria for different grade levels assume that the same (basic) skills are sufficient for ELL student success in the mainstream classroom. Language proficiency needed for successful participation in a Kindergarten classroom clearly differs, however, from that required for a high school mainstream classroom or even a third grade classroom (Corson, 2001). As a result of such policies, a mismatch between the skills developed in the ESL program and those required in the mainstream classroom can emerge (Chamot and O’Malley, 1994; Lucas and Wagner, 1999). This, in turn, may lead to inappropriate referrals to special education services (Cummins, 1984; Gersten, 1996). Length Of Time In The Program The length of time ELLs participate in a bilingual/ESL program has also received much attention, particularly as an indicator of program effectiveness. Gersten and Woodward (1995) state that, “a critical indicator of the success of any bilingual instruction program for language minority students is the rate at which student leave specialized classes for second language learners and enter mainstream classes” (p. 227). Some have claimed that prolonged segregation from mainstream peers will hinder ELLs’ English language development and academic progress (Porter, 1998; Rossell, 2000). There is little evidence, however, that bilingual students “languish” in their programs (Krashen, 2001). Several states indicate that their average program participation of ELLs is around three years. For instance, the Florida Department of Education reported that an average length of program participation of 3.1 in 2000 across all grade levels., with secondary students (grades 6 through 10) remaining slightly longer (four to five years; OMSLE, 2001). Illinois reported that over 77% of the students in bilingual programs had been in the program for less than three years (Illinois State Board of Education, 2000). Regardless of these trends, extended program attendance continues to be perceived as a negative feature of a bilingual/ESL program, regardless of its design. Without examining exited students’ academic performance after they have exited, it is difficult to draw conclusions about the role of length of stay in the program. Length of stay in the program is also interpreted to imply that the more quickly ELLs exit, the better the program (Gandara and Merino, 1993; Linquanti, 2001). The assumption that a direct 2 In some cases, the discussion is reduced to exit percentages. For instance, Ron Unz, the founder of Proposition 227 to abolish bilingual education in California, criticized bilingual education for having a “95% failure rate” because it was reported that only 5% of ELLs were reclassified annually. After exit: Academic achievement patterns of former English language learners 4 4 relationship exists between exit rates and program success ignores, however, that the rate of student progress through a bili


Introduction
The inclusion of English language learners (ELLs) in recent accountability and school reform efforts (e.g., No Child Left Behind) has placed bilingual and English as a Second Language (ESL) programs under renewed scrutiny as to their ability to overcome the linguistic barriers ELLs encounter when entering U.S. schools.With few exceptions, the effectiveness of bilingual/ESL programs has been evaluated on the performance of ELLs still being served through the program.
While providing some information about the performance of limited English proficient students, these program evaluations are unable to describe what happens to ELLs once the latter have been placed in a mainstream classroom on a fulltime basis after having been reclassified as "fluent English proficient."This article is concerned with the question whether ELLs in fact "catch up" academically after they exited from a bilingual or ESL program.It is argued that including former ELLs in accountability planning will provide a better understanding of whether a school or district has been able to provide equal educational opportunities for this group of students.After a brief overview of the schooling of ELLs in the United States, the article discusses exit policies, the role of exit rates in ELL program evaluations, and academic achievement patterns for former ELLs.Then, the results of a study examining former ELLs' 4 th and 8 th grade achievement patterns in English language arts, mathematics, and science will be presented.The article concludes with the study's implications for future program evaluations and policies.

Issues of Exit in ELL Programs
Background Title VI of Civil Rights Act (1964) requires school districts to take affirmative steps to avoid discrimination against ELLs based on their limited proficiency in English.In the landmark court case Lau v. Nichols (1974) the Supreme Court proclaimed, "there is no equality of treatment merely by providing students with the same facilities, textbooks, teachers, and curriculum; for students who do not understand English are effectively foreclosed from any meaningful education" (cited in Teitelbaum and Hiller, 1977, p. 7).The court refrained, however, from recommending one particular program to remedy such discrimination, pointing out that the intervention could be instruction in the students' native language or in English.In Castañeda v. Pickard (1981) the court identified three criteria to assess when a district's intervention can be considered adequate: it must have theoretical merit, it must be implemented with sufficient resources, and it must demonstrate its effectiveness through program evaluation (Crawford, 1999).
Not surprisingly then, different programs have been developed and implemented to meet the needs of ELLs over the last three decades.Generally speaking, these programs differ according to the language of instruction (bilingual or English-only), program goals (bilingualism or English proficiency), and the program's target group (ELLs only or ELLs and native English speakers) (Brisk 1998;Ovando, Collier, and Coombs, 2003).Much of the controversy regarding the schooling of ELLs has focused on whether schools should adopt bilingual or English-only approaches (for detailed overviews of the effectiveness debate, see August and Hakuta, 1997;Collier, 1992;Crawford, 1997;Cummins, 1996;Rossell and Baker, 1996).One implied feature shared by the majority of the programs is their temporary nature: the ultimate goal is to prepare ELLs for placement in a mainstream classroom within a certain time frame.As a result, deciding when and how to exit ELLs from a bilingual or ESL program plays an integral role in the schooling of ELLs. 1 Furthermore, the temporary nature draws attention to the length of time the students remains in a program.Both issues are discussed in more detail below.

Exit Policies
Little agreement exists as to how to determine eligibility of minority language speakers for specialized language services (e.g., Rossell and Baker, 1988) and, subsequently, on how to decide that ELLs no longer need bilingual or ESL services.There is substantial variation within and across states in exit policies and guidelines.States and districts differ widely in whether to use a cut-off score on a test and in what cut-off score to select to determine exit-readiness (De Avila, 1990).Further, some states have no guidelines regarding which assessments to use, others either recommend or mandate the use of certain instruments, including standardized achievement tests, English oral language proficiency tests, English literacy tests, or teacher judgment (De George, 1988;Gandara andMerino, 1993, Fleischman andHopstock, 1993;Rossell and Baker, 1988).Comparison of exited students across states and even across districts is therefore extremely difficult (Linquanti, 2001).
The choice of assessment instrument, the use of cut-off scores, and the data used to make placement decisions for ELLs reflect a district's definition of the linguistic needs of ELLs, the purpose of the ELL program, and expectations for mainstream classroom teachers to be able to work with ELLs (DeAvila, 1990;Nadeau and Miramontes, 1988).For instance, defining ELLs' needed language proficiency as survival oral language proficiency may translate into an exclusive reliance on oral assessments for exiting and a focus in the ESL program on oral language development rather than literacy development.Similarly, exit policies that use similar criteria for different grade levels assume that the same (basic) skills are sufficient for ELL student success in the mainstream classroom.Language proficiency needed for successful participation in a Kindergarten classroom clearly differs, however, from that required for a high school mainstream classroom or even a third grade classroom (Corson, 2001).As a result of such policies, a mismatch between the skills developed in the ESL program and those required in the mainstream classroom can emerge (Chamot and O'Malley, 1994;Lucas and Wagner, 1999).This, in turn, may lead to inappropriate referrals to special education services (Cummins, 1984;Gersten, 1996).

Length Of Time In The Program
The length of time ELLs participate in a bilingual/ESL program has also received much attention, particularly as an indicator of program effectiveness.Gersten and Woodward (1995) state that, "a critical indicator of the success of any bilingual instruction program for language minority students is the rate at which student leave specialized classes for second language learners and enter mainstream classes" (p.227).Some have claimed that prolonged segregation from mainstream peers will hinder ELLs' English language development and academic progress (Porter, 1998;Rossell, 2000).There is little evidence, however, that bilingual students "languish" in their programs (Krashen, 2001).Several states indicate that their average program participation of ELLs is around three years.For instance, the Florida Department of Education reported that an average length of program participation of 3.1 in 2000 across all grade levels., with secondary students (grades 6 through 10) remaining slightly longer (four to five years; OMSLE, 2001).Illinois reported that over 77% of the students in bilingual programs had been in the program for less than three years (Illinois State Board of Education, 2000).Regardless of these trends, extended program attendance continues to be perceived as a negative feature of a bilingual/ESL program, regardless of its design.Without examining exited students' academic performance after they have exited, it is difficult to draw conclusions about the role of length of stay in the program.
Length of stay in the program is also interpreted to imply that the more quickly ELLs exit, the better the program (Gandara and Merino, 1993;Linquanti, 2001). 2 The assumption that a direct 4 relationship exists between exit rates and program success ignores, however, that the rate of student progress through a bilingual or ESL program will necessarily be influenced by individual student variables, program quality, as well as school variables (e.g., Gandara and Merino, 1993;Thomas and Collier, 2002).If exit rates are to be used as an outcome measure, they cannot be solely tied to a particular program but must include a wide range of contextual variables.Moreover, as was illustrated above, inconsistent exit practices make it difficult to draw conclusions about a program's ability to exit students or to compare districts or states.A district with a low exit cut-off score may have a higher exit rate than a district with a more challenging cut-off score.In this case, comparing each district's exit rate would not reveal anything about the quality of the program.
Second, assuming "mainstreaming minority children out of a bilingual program into an English-only program will promote the development of English literacy skills more effectively than if children were maintained in a bilingual program" (p.129-130) has also been criticized.Cummins (1980) labels this assumption the "exit fallacy," warning that a premature exit from a bilingual or ESL program may negatively affect ELLs' academic success in the mainstream.To be sure, current second language (L2) acquisition theory emphasizes interaction with native English speakers for effective L2 development (e.g., Ellis, 1999;Fillmore, 1991).Further, student segregation may lead to the marginalization of the schooling of ELLs (De Jong, 1996;Valdés, 2001).Yet, there is little empirical evidence that placement in a mainstream classroom will necessarily be more beneficial for ELLs.For instance, ELLs who are placed in mainstream classes without specialized services (due to parent waiving access to such services) perform well below the national norm on standardized tests (Thomas and Collier 2002;Waters, 2001).Several studies (Harklau, 1999;Harper and Platt, 1998;Valdés, 2001;Verplaetse, 2000) have documented the lack of ELL classroom participation and the absence of interactions between ELLs and native English speakers in secondary mainstream classrooms.A lack of comprehensible input, limited access to academically engaging instruction, and few opportunities for second language development can make mainstream classrooms ineffective learning environments for ELLs.
Finally, it is not empirically established that a quick exit implies long-term academic school success, when controlling for student background and program variables.Indirect evidence against this "quick-exit" assumption comes from studies showing that ELLs in long-term bilingual programs are successful academically (e.g., Holm and Holm, 1990;Lindholm-Leary, 2001;Ramirez, 1992;Thomas and Collier, 2002).Ramirez (1998) found that former students performed better than English speakers on standardized tests in the San Francisco school district and reports that these students had, on average, attended a bilingual program for 4-5 years.On the other hand, Gersten and Woodward (1995) report that after 3-4 years of program attendance, former ELLs still performed under the 25 th percentile in middle school on the Comprehensive Test of Basic Skills (CTBS).Clearly, long program attendance does not necessarily impede nor guarantee academic success.
The relationship between exit rate and subsequent school success was more directly explored in a study the New York Department of Education (2000).The study found that ELLs who exited in the early elementary grades (K-4) and who entered in PK/K-2 performed at or above the citywide average in Math and Reading in 7 th and 8 th grade, respectively.This study suggests that perhaps entry grade and exit grade level may play a role in predicting ELLs school success after exit (see also Goldberg, 1997).
Though faulty in its reasoning (e.g., De Jong and Ramos, 2003), the California debate illustrates the dominance of this issue in setting educational policy.See also Grissom (2004) for a longitudinal analysis of exit percentages.

Former ELLs and Program Accountability
Proponents and opponents of bilingual or ELS approaches have used program evaluations to argue for or against the implementation of certain programs.Labeled "advocacy-oriented" program evaluations by August and Hakuta (1997), these studies typically compare the outcomes of a bilingual and an ESL program in the first three years of implementation.Though severely criticized on methodological and theoretical grounds (August and Hakuta, 1997;Rossell and Baker, 1996;Willig and Ramirez, 1993), these studies nevertheless continue to fuel much of the debate on the schooling for ELLs.Importantly, these evaluations have typically focused on the achievement patterns of ELLs as they are still classified as limited English proficient.While such an approach can provide useful short-term feedback, it cannot assess whether the program has succeeded in preparing the students for effective participation in the mainstream.To answer this question, program evaluations must go beyond the bilingual/ESL program achievement data and include data on the school performance of "graduates" from such specialized programs.
Few studies have explored the academic achievement of "former ELLs."ELLs are rarely tracked once they have exited from the bilingual/ESL program and few states disaggregate and report their statewide testing data for former ELLs.An examination of the achievement patterns of two states with high ELL enrollment that do report such data, Florida and California, illustrates that former ELLs tend to lag behind fluent English-speaking peers, particularly at the secondary level.In Florida, for example, the Florida Department of Education (2001) reports that between 12 and 20 percent fewer former ELLs pass the state's test than fluent English speakers in math and reading at grades 4, 5, 8, and 10.Second, former ELLs tend to perform better on math tests than on reading or content area (science, social studies) tests.SAT-9 data from California show that about 10-12% fewer former ELLs score above the 50% percentile in science and social studies (Grades 9-11), whereas their scores for math at these grade levels essentially parallel those of English-only students (California Department of Education, 2001).These statewide achievement patterns underscore the need to include former ELLs in state and district accountability systems.
The exclusion of former ELLs from program and district accountability reflects the perception that the responsibility for meeting the needs of ELLs is that of the bilingual/ESL teacher.Once a student is "fixed" and exited from the program, mainstream teachers (expect to be able to) treat them as native English speakers (Cummins, 2000;Lucas and Wagner, 1999;Shannon, 1990).Thus, from a mainstream perspective, exited students are not distinguished as a separate group because they are now "native English speakers."They are also largely ignored from a bilingual/ESL perspective because they no longer fall within the confines of specialized services.In contrast, the inclusion of former ELLs in the accountability debate of bilingual/ESL programs acknowledges that the entire school/district is responsible for the learning of ELLs and not only the bilingual program (Carter and Chatfield, 1986).It also emphasizes that this accountability includes a second language learner's entire school career and not only the few years that s/he attends a bilingual/ESL program.
The purpose of this study was, therefore, to examine the academic achievement of former ELLs.For the purpose of the study, former ELLs are defined as those students who were classified as limited English proficiency upon school entry and attended a bilingual or an ESL program, who were then reclassified as fluent English proficient, and enrolled fulltime in a mainstream classroom.In addition to general achievement patterns in English language arts, math, and science, the study also explored role that length of bilingual/ESL program attendance played in predicting academic 6 success.To examine the role of varying academic proficiency demand from grade to grade, the analysis also included the grade at which students exited as a second variable.

Research Design
The study took place in medium-sized school district in the Northeastern United States.The district enrolls close to 8,000 students and implements a bilingual (BE) and an English as a Second Language (ESL) program.The district was chosen because of its long-standing experience with ELLs (the first bilingual classroom was established in the late 1960s), its consistency in program philosophy and implementation, and its emphasis on quality of education for all ELLs.The study explores the following two research questions.The first question is whether former ELLs in this district close the academic achievement gap with students in regular education.The second research question examines the role of two variables (length of program participation and grade of exit) in predicting academic achievement.

Program context
The district's bilingual education (BE) program is designed for native Portuguese or Spanish speakers with limited English proficiency and who are dominant in their native language.The goal of the program is to prepare its students to succeed in a mainstream classroom.Prior to 1995, the elementary bilingual program was run as a traditional transitional bilingual education program with a focus on transitioning to English literacy and exit within three years.After 1995, the model changed to include continued access to the native language and an exit process that emphasized readiness for the mainstream classroom.When considering the data below, it should therefore be kept in mind that the 8 th grade former BE students went through an early-exit bilingual program design at the elementary level, whereas the 4 th grade former BE students attended a program that was in line with the new model, i.e., a late-exit/maintenance program.The elementary BE programs are located in schools with the largest proportions of students eligible for free/reduced lunch.The Portuguese programs are located in two schools where more than 50% of the students are eligible free/reduced lunch (63% and 51%).The two Spanish bilingual programs have 38% and 29% of their students eligible for free/reduced lunch, respectively.
The ESL program is a self-contained program with the same objective as the BE program: to prepare students to succeed in a mainstream classroom.The elementary ESL program is taught in English and serves a multilingual population.Students' home language backgrounds include Russian, Chinese, Hindi, Japanese, Korean, and several other languages.The school in which the ESL program is located has the lowest percentage of students eligible for free/reduced lunch (16%) in the district.Traditionally, the ESL program has served a student population with access to educated parents and native-language resources outside school, such as a Korean Saturday school.

Exit guidelines
The process of exiting students is the same for the ESL and the bilingual program and is outlined in a district document.For oral language, students are expected to score minimally a 4 on comprehension and on production on a rating rubric that ranges from 0 (no oral proficiency) to 5 (oral proficiency is like a native speaker).For English literacy, students are expected to score a Level 3 (fluent reader/writer) on the Language Assessment Scale, a formal language proficiency test (Duncan and De Avila, 1988).Academic performance is not included in the form of formal tests.Bilingual and ESL teachers understand, however, the limitations of formal tests and take a wholechild approach when making decisions about exiting students, including student personality and attitudes as well as strengths in the content areas.

Participants
The sample of exited students was determined by examining exit forms used by the district since 1995 as well as individual student records in the bilingual department.Students whose parents requested their child's exit before the teacher recommended such an exit were excluded as were exited students for whom it could be determined that they were eligible for special education services.This resulted in a total sample of 560 identified exited students who were recommended for exit during or after Kindergarten.To answer the research questions, a subset was created for those exited students who took the state test between 1998 and 2000 in grade 4 or grade 8.This process resulted in a sample of thirty-eight 4 th grade exited students (twelve from the BE program and twenty-six from the ESL program) and fifty-six 8 th grade exited students (thirty-six BE program students and twenty ESL program students) for whom all required data were available (see Table 1).The majority of the ESL program students exited between Kindergarten and second grade (76%); 13% exited between third and fifth grade and between sixth and eighth grade.Of the exited BE program students, one-third exited before third grade, 60% exited between grades three and five and 7% exited during the middle school years (grade 6-8).The average length of program participation for the 4 th grade ESL program students was 2.2 years (median = 2.0; range =3.6) and for the BE program students 3.1 years (median=3.0;range 2.7).The 8 th grade ESL program students stayed in their program for an average of 2.6 years (median=2.0;range=6.0),whereas the BE program students completed an average of 3.4 years (median=3.0;range=5.0).The exit figures parallel those in the state where more than 80% of the ELLs are reported to exit within three years.

Instruments
The academic achievement outcome measure is a mandated state test developed by the state based on its curriculum standards.Between 1998 and 2000 the state test was administered for English Language Arts (ELA), Mathematics, and Science at grades 4, 8, and 10.The test consists of multiple choice, short-answer, and essay questions, and also includes a writing sample for English Language Arts.Students' raw scores are converted into scaled scores, which are then divided into four proficiency categories: Failing, Needs Improvement, Proficient, and Advanced Proficiency.The study examined distribution of 4 th and 8 th grade exited and "regular education students" over the proficiency levels in the three subject areas (ELA, Mathematics, and Science)."Regular education" excludes students classified as ELLs as well as students eligible for special needs services.

Analysis
To answer the first research question, whether exited ELLs are performing at similar levels as their peers, "closing the achievement gap" was defined as exited students showing similar patterns on the state test as students in regular education in the district and/or the state as represented by the percentage of students in each of the proficiency categories.Because there were so few students in the Advanced category, it was decided to collapse the Proficient and Advanced Proficient categories into one.Both categories indicate that students are meeting state expectations for the grade level.The analysis therefore considered the percentage of exited and regular education students in ELA, Mathematics, and Science at three proficiency levels: Failing (F), Needs Improvement (NI), and Proficient (P).
The second question examines the role of time variables.The first variable is the variable related to exit rates, namely Length of BE or ESL program attendance.This variable was defined as the period between entry into the bilingual/ESL program and the date of exit (one school year (9 months) is one year).Secondly, the analysis considered the exit grade: the grade level at which was the student exited (e.g., second grade or fourth grade) to take into consideration the increased levels of academic demand placed on former ELLs as they exit.A regression analysis was done to see which, if any, variables had a significant effect on achievement outcomes for each of the achievement tests (ELA, Mathematics, and Science).While these two programs are in the same district, direct comparison between the two programs is not possible since student background variables were not controlled for.

Academic achievement patterns
To answer the question whether exited students close the achievement gap when compared to the regular education population, the percentage of students scoring in the Failing (F), Needs Improvement (NI) or the Proficient (P) proficiency levels was calculated for 4 th and 8 th grade exited students for three subject areas, ELA, Mathematics, and Science (Tables 2 and 3).T-test analysis revealed significant program differences for 4 th grade mathematics (p<0.01)and, to a lesser extent, for 4 th grade science (p=.04), and 8 th grade math and science (p=0.04 and p=0.03, respectively).No significant differences were found for English language arts between the ESL and BE program.The results will therefore be presented by program model.

ESL program results
The 4 th grade ESL exited students and the district's regular education students show a similar distribution of scores in the three subject areas (Table 2).However, fewer ESL program students score at the Proficient level in ELA as compared to regular education students.
The 8 th grade ESL exited students show similar achievement patterns as students in regular education at the district and the state level for ELA and mathematics.For science, on the other hand, more ESL exited students' scores fall into the failing category, although they match the district and the state patterns at the proficient/advanced proficiency level.Like the elementary students, fewer 8 th grade ESL exited students scored at the Proficiency level in ELA.

Bilingual Program Results
More BE exited students score in the Needs Improvement category than in the Proficient category for mathematics and science, a pattern different from the regular education students.The BE exited students more closely parallel the distribution of the regular education students at the district and state level for ELA (see Table 3).
The 8 th grade exited BE students show a different pattern.Their score distribution diverges from regular education students, having more students in the failing category for mathematics and science and fewer in the Proficiency category.For ELA, on the other hand, exited BE program students resemble the pattern for regular education students.Like the elementary students, fewer BE students scored at the Proficiency level in ELA.

Length Of Program Participation And Grade Exited
The second research question explored whether two variables, length of program participation and grade exited, predicted exited students' academic achievement outcomes.No significant correlations were found between length of program participation and scaled scores.When controlling for program model, the regression analysis revealed the following patterns.Exit grade level emerged as a significant predictor for 4 th grade ELA and Science (p<.01), i.e., the higher the grade level that the student exited, the lower their scores on the 4 th grade state test in ELA and Science.At the 8 th grade level, neither variable was found as a significant predictor of academic achievement.A significant interaction effect between grade exited and program model emerged, however, for 8 th grade science (p<.09), though with a small effect size (r=.05).For the exited ESL program students, the lower the grade level they exited, the better their 8 th grade science scores.Exited BE program students' scores, on the other hand, slightly improved as they exited at higher grade levels.

Discussion
This study considered the achievement of bilingual students who at one time were classified as "limited English proficient", attended a bilingual or ESL program, and who are now participating in a mainstream classroom without specialized services.The first research question examined the extent to which former ELLs in one district were able to meet grade level expectations as compared to their peers.Results indicate that, in general, 4 th grade exited students' score distribution over the three proficiency levels (Failing, Needs Improvement, Proficient) resembled those of regular education students at the district and state level for ELA, Mathematics, and Science.At the same time, more former BE and ESL program students scored in the Needs Improvement rather than the Proficient category for ELA.The same trend emerged for mathematics and science for exited BE program students.For the elementary school level, then, the well-designed BE program and ESL program, together with the mainstream program, appear to meet the language and academic needs of ELLs when considering proficiency categories.
The fact that fewer 4 th grade exited students are "proficient" raises an important question for the future: will these former ELLs eventually perform at the proficient level or will the gap widen as more demands are placed on students' ability to mediate content through language?Non-cohort standardized state test data from Florida and California suggest a widening rather than a closing gap between regular education students and former ELLs for reading as well as the content areas for students at the higher grade levels (California Department of Education, 2001;OMSLE, 2001).On the other hand, quasi-longitudinal data from bilingual programs suggest that ELL may close the gap over time, provided they have had a strong bilingual schooling base for more than 4 years (Thomas and Collier, 2002).These data emphasize the importance for school districts to track former ELLs throughout their school career.This can be achieved when districts keep a "language status" code for ELLs when these students leave their program.Most districts identify ELLs through a code such as ELL or LEP (limited English proficient) but delete such identification when the student leaves the ESL/bilingual program.Instead, the ELL/LEP code can be changed to a code that identifies them as a mainstreamed student.For example, the Florida Department of Education uses the code "LF" for students who exited within the last two years and "LZ" for students who exited more than two years ago.Such a code could also contain program type data.Long-term cohort data for former ELLs is needed to see if a gap exists and whether it widens, closes, or remains the same as former ELLs go through the grades.
The analysis of the 8 th grade exited students' academic achievement showed a different picture.The gap between former exited students and the regular education students was the largest for mathematics (BE exited students only) and science (both BE and ESL exited students) with more students' scores falling into the Failing category.Counter to other district-level studies (e.g., Abella, 1992;Gonzalez, 2001;Waters, 2001), this study found that the former ELLs largely matched the score distribution of English speakers over the three proficiency levels for English language arts.This may be partially due to a strong middle school English department in the district.It may also be related to former ELLs' access to acquiring the necessary English skills during their elementary schooling (cf. the stronger ELA patterns for the 4 th grade former ELLs).This finding also underscores the importance of distinguishing between the language skills developed through ELA and the content areas.The content areas have their own register and make specific language demands on students, particularly at the secondary level (e.g., Carrasquillo and Rodriguez, 1996;Corson, 2001;Short, 2002), and language skills learned in the ELA class may not necessarily be sufficient or appropriate for the content area classroom.Even very fluent bilingual students may need teachers to provide them with more sophisticated ways to discuss and engage with academic content (Cummins, 2000;Freeman, Freeman, and Mercuri, 2002;Gibbons, 2002) and demonstrate their learning on a formal test.
The 8 th grade patterns also revealed a gap between BE and ESL program students, with the ESL program students performing better than the BE students in math and science.Several factors may contribute to these program differences.First, the elementary ESL program is located in a school with a higher socio-economic status and has traditionally enrolled students from homes with higher parent education levels.The 8 th grade students would have been representative of this group of ESL program students.The BE program students, on the other hand, are primarily from lower socio-economic background and their schools have an overall higher percentage of students who qualify for free/reduced lunch.While the BE students had remained in their program slightly longer (3.4 years versus 2.6 years for the ESL program students), this difference was statistically insignificant (see Krashen, 1996, however, regarding the role of socioeconomic status and program length).Another factor that may have played an important role is the nature of the elementary bilingual program that the 8 th grade BE students had attended.They went through an early-exit bilingual program that focused on English literacy development with a quick shift to English in the early grades.If the bilingual students exited the bilingual program proficient in English but were lagging behind academically, it would have been difficult to catch up in these areas.ESL program students, on the other hand, were exposed to a full curriculum in the mainstream earlier on.Gradelevel content area development in longer-term programs, such as a bilingual education program, is therefore crucial.While earlier test data are unavailable for the 8 th graders, this study's 4 th grade BE program results suggest that the revised elementary BE program (a late-exit bilingual program) may be providing ELLs with a better start in the mainstream classroom in English language arts as well as math and science.This trend would be consistent with other studies in the field (e.g., Thomas and Collier, 2002).
The third research question examined the role of two variables: length of program participation and grade level exited.When controlling for program type, length of program participation was not a significant predictor of exited students' achievement patterns.While it is often assumed that students who stay in a program longer are students with academic difficulties, this study found no significant correlation between length of time in the program and academic achievement scores in either program.Exit grade, on the other hand, did emerge as a significant variable, at least for the 8 th grade students.This finding may reflect the fact that the language of school increasingly becomes more dependent on students' ability to negotiate meaning through linguistic means (Cummins, 2000;Corson, 2001).Students move from predominantly oral discussions and activities grounded in their own personal experiences (K-2) to writing and talking about abstract content-related concepts and they must master the more complex vocabulary, syntactic structures, and pragmatic conventions that are appropriate for the specific discourse of the subject and the grade level (de Jong & Harper, in press).Guidelines for exit should include this awareness of students' ability to manipulate language for academic purposes appropriate to the grade level they are entering.Bunch, Valdez, Lotan, Cohen, and Abram (2002), for instance, developed an academic oral language assessment that uses grade-level appropriate academic classroom tasks that require students to explain and persuade.Rather than a fixed criterion for all grade levels, a more flexible approach is needed that considers the increased demands that will be placed on an exited student and the extent to which continuing academic language needs of ELLs can be met in the mainstream classroom.Instead of policies that label ELLs either as "in" or "out" of specialized services, a more gradual approach would allow services to change as ELLs become more proficient (Gandara and Merino, 1993;Linquanti, 2001).Moreover, it stresses the importance for mainstream teachers to be prepared to continue the development of the academic language proficiency once ELLs exit (Cummins, 2000;Echevarria, Vogt, and Short, 2000;Gibbons, 2002).
Finally, the significant interaction between exit grade and program found for 8 th grade exited students is intriguing, though the results must be interpreted with caution due to the small sample size.The ESL program did better if students entered exited in the lower grades, whereas students in the BE program did better if they exited at a higher grade.In the absence of a significant correlation between length of program attendance and achievement, this finding points to the possibility of a differential impact of program type, possibly in combination with socioeconomic status.If the ESL program closes the gap for ELLs with more affluent parents and who exit at lower grade levels (i.e., K-2) but not for students who exit at higher grade levels and/or whose needs are more extensive than can be met within a short time, its effectiveness is limited to a very specific target population.A similar differential program effect was observed in a study by the New York Board of Education (2000).In this study, the achievement patterns for students who had attended a bilingual program were less sensitive to length of program attendance than those of ESL program students.Such a finding supports the need to move away from a one-model-fits-all approach to more complex ways to meeting the needs of ELLs.Clearly, more research is needed for a better understanding of the interaction between program features, school variables, student characteristics, and academic achievement patterns (August and Hakuta, 1997;Nadeau and Miramontes, 1988).Although the analysis has limited power due to the small number of students at each grade level, the role of exit grade warrants further exploration with a larger sample.

Conclusion
Exited ELLs are important to consider within a larger framework of district accountability.The analysis of former ELLs' achievement data illustrates that former ELLs may not necessarily catch up.This is particularly true for the secondary level, but may remain hidden at the elementary level as students "pass" the test at the lowest level, but are not quite proficient yet.These findings, which are also found in states such as California and Florida, underscore that the schooling of minority language students is not only the responsibility of bilingual or ESL teachers but continues 13 after students have been exited from such programs.Statements about achievement patterns of ELLs and native English-speaking students or claims that linguistic barriers for ELLs have been overcome can only be examined when exited ELLs are included in the analysis.A district cannot be said to have met ELLs' academic and linguistic needs if discrepancies in academic achievement patterns between exited ELLs and native English speakers persist.It is therefore important to disaggregate data for former ELLs and follow their achievement over time.
When program evaluations include former ELLs' achievement, several issues emerge that traditional advocacy-oriented program evaluations, that only include students who are still labeled as "limited English proficient," tend to overlook.By examining the role of length of program participation and exit grade, this study confirmed that length of program participation does not appear to play a significant role in predicting future student success.Furthermore, it raised questions regarding the relationship between exit grade and ELLs' subsequent academic achievement, the appropriateness of specialized programs for students with certain background characteristics and ages and emphasized the importance of examining long-term academic achievement patterns (Thomas and Collier, 1997).It will not be sufficient, however, to only examine academic outcome data.The results in this study for the ESL and bilingual program, for 4 th and 8 th graders, and across subject matters also raise important questions regarding the factors that facilitate or hinder the exit process, how students and teacher mediate the exit process, as well as the social, linguistic, and academic needs of exited students when they first encounter the mainstream classroom and continuing over time as language and content demands increase (e.g., Franson, 1999;Shannon, 1990).
Including exited students or former ELLs in the analysis approaches district accountability from a broader perspective than previously taken in advocacy-based program evaluations.Meeting the specialized needs of ELLs cannot be the sole responsibility of a bilingual or ESL program.As Mora (2002) points out, such program attendance represents, on average, an approximate 4% of a student's school career.A focus on former ELLs emphasizes accountability for all teachers involved with bilingual students throughout the latter's school career.