English-Language Assistance Programs

English-Skill Acquisition, and the Academic Progress of High School Language Minority Students; Statistical Data Included

Utilizing the 1988-1992 surveys from the Restricted-Use National Education Longitudinal Study of 1988, I find that language minority students who received English-language assistance (ELA) programs beyond the third grade reported lower English-skill acquisition and made smaller academic progress in high school than their peers. Moreover, while the type of high school ELA program (e.g., bilingual education) differently affected scholastic outcomes, students in such programs did not surpass their counterparts in monolingual-English classes on average.

Schooling programs designed for language minority students have been facing growing political, social, and economic criticism in the United States. The June 1998 passage of California’s Proposition 227, which eliminated most of the state’s bilingual education programs in favor of sheltered one-year English-immersion methods, exemplifies the general public’s increased skepticism over the current educational methods used to teach limited-English-proficient (LEP) children in this country. [1]

Although LEP students represent at least 5% of all public school children in grades K-12 (Han, Baker, Rodriguez, & Quinn, 1997), the effectiveness of current English-language assistance (ELA) schooling programs, such as bilingual education, lacks empirical consensus. While much has been written on this topic, many studies rely on qualitative outcomes (such as teachers’ perceptions), use simple descriptive statistics, or analyze students within one school or district. [2] As noted by Tucker (1980, p. 2) two decades ago, “despite more than 10 years of federal support for bilingual education, very little critical, empirical, or longitudinal research has been conducted” [in the U.S.]. More recently, Glenn (1997, p. 2) observed “… something like $ 100 million has been spent on research and evaluation of bilingual education over the last 30 years, with remarkably little in the way of bankable results that have been applied to practice.”

Insight into some potential long-term effects of ELA programs may be inferred from their influence on English-skill acquisition and scholastic progress. [3] For example, the extent to which these programs directly or indirectly influence English proficiency could affect future labor market opportunities. [4] Scholastic progress associated with ELA programs also may alter students’ decisions to enter certain fields or to acquire higher education, which in turn could impact socioeconomic outcomes particularly in light of the increasing returns to education (e.g., Murphy & Welch, 1992).

In what follows, I use 1988-1992 panel data from the Restricted-Use National Education Longitudinal Study of 1988 (NELS:88) to analyze empirically whether ELA programs affect the English-skill acquisition, academic progress, and school attrition of language minority youths. My analyses reveal that language minority students in NELS:88 who received some form of ELA after the third grade on average acquired less English fluency and made lower academic progress (measured by growth in cognitive exam scores) than their otherwise similar counterparts. The results further suggest that the type of high school ELA (such as bilingual education versus English-as-a-Second-Language) differently affected these outcomes, although in all cases students in specific ELA programs did not surpass their peers.

Background and Overview of ELA Programs

Policies designed to guide the education of language minority children have been established at the federal, state, and local levels. For example, the U.S. Supreme Court ratified in Lau vs. Nichols (1974) that school districts failing to provide understandable instruction obstruct LEP students’ rights to equal educational opportunity (Donato, Menchaca & Valencia, 1991; Meier & Stewart, 1991). Many times, the final decisions on which schooling methods to implement for LEP students depend on school administrators, creating a lack of conformity of ELA programs between states and even between schools in the same district. [5]

Mainstream ELA programs include bilingual education (BE) and English-as-a-Second-Language (ESL). The underlying concept of BE involves both the minority and majority languages during instruction. The majority of BE programs in this country are transitional, in which the use of the minority language as an educational tool diminishes over time as students learn English. [6] Students in ESL attend mainstream classes for part of the day, and receive formal English instruction at other times of the day. While many ELA programs tend to be implemented during early grades, high school ELA programs remain important because they represent the final formal education received by many LEP students before entering the work force.

Supporting Evidence in Favor and Against ELA Programs

Individuals in favor of BE assert that instruction in the two languages allows for cognitive skill development in the minority language; these skills eventually transfer to the English-speaking domain of the student. Spanish reading skills, for example, should transfer into English reading capability (e.g., Crawford, 1995; Ovando, 1983). Moreover, some supporting research suggests that students who participate in BE programs have relatively higher exam scores, higher daily attendance rates, and lower dropout rates than LEP students immersed in monolingual-English classes (e.g., Greene, 1998; Willig, 1985). Supporters of BE also contend that monolingual-English instruction denies LEP students an opportunity for equal education (e.g., the Lau case). Other scholars claim that language minority students have the right to retain their native languages in instruction (e.g., Rodgers, 1995).

Not all studies provide positive evidence. Recent work finds that first and second generation Hispanic immigrants who received ELA at some point during their education earned significantly less ten years after high school than their peers (Lopez & Mora, 1998). Other studies suggest that “structured” English-immersion is more effective than BE (Rossell & Baker, 1996; Rossell & Ross, 1986). Additional research notes a lack of evidence for either long-term advantages or disadvantages associated with ELA programs versus English-immersion (e.g., August & Hakuta, 1997; Baker & de Kanter, 1981; Danoff, 1978).

The majority of work analyzing ELA programs involves elementary school students, although a few exceptions exist. For example, Curiel, Rosenthal and Richek (1986) find some evidence in favor of BE at the junior high and high school levels. Conversely, Ho (1985) fails to uncover empirical advantages for native-language instruction versus English-immersion in six academic subjects but one (economics/public affairs) for eighth graders. Finally, Lopez (1998) notes the relationship between ELA taken in later grades and educational attainment at best seems to be statistically insignificant.

Current Criticisms of ELA Programs

Many recent criticisms stem from the perception that ELA programs are not effective in practice, regardless of the pedagogy. At least five distinct criticisms appear in the debate over ELA programs, which may explain some of the inconclusive empirical evidence. If these criticisms accurately reflect the current status of ELA, an overhaul of these programs may be necessary to ensure equal education opportunities for language minority students.

One criticism of ELA programs relates to segregation, despite the 1954 ruling by the U.S. Supreme Court outlawing public school segregation on the basis of race, religion, or national origin in Brown v. Board of Education (e.g., Donato, Menchaca & Valencia, 1991; Meier & Stewart, 1991). Some researchers have noted that this segregation decreases the likelihood of assimilating into English-oriented environments (Meier & Stewart, 1991; Porter, 1996), and involves the potential lack of “separate but equal” education opportunities. Indeed, ELA at times has been used to segregate Hispanic LEP children into lower-quality classrooms or underfinanced schools (e.g., Donato, Menchaca & Valencia, 1991).

A second criticism stems from the lack of appropriate training of many teachers and aides in ELA programs. Less than 3% of all teachers instructing LEP students had obtained an academic degree with a specialization in BE or ESL in the early 1990s, and only 30% of teachers with LEP students had received any form of LEP-instruction training (Han et al., 1997). If teachers do not possess the necessary training, then LEP students in ELA may be less likely to academically succeed.

Third, cases have been documented in which LEP children receive BE in languages outside of the students’ domain. [7] For example, in a memo to the superintendent of the Albuquerque Public School (APS) District, the Department of Education Office for Civil Rights (OCR) asserted that APS placed some Native American and Asian LEP students into Spanish-English bilingual classes (Chavez, 1995). Headden (1995) notes a similar case in which Chinese-speaking students in San Francisco had been enrolled in Spanish-English courses.

A fourth criticism of ELA programs involves academic “tracking”, in which some language minority students potentially learn remedial coursework, as discussed by Meier & Stewart (1991). If the class material has been “slowed down” compared to the mainstream, it becomes much harder to exit ELA or move forward scholastically because the students have not been taught the necessary academic skills.

Finally, in some cases children have been enrolled or retained in transitional BE programs despite being proficient in English (San Miguel, 1984). One possible reason for the placement concerns additional funding schools receive from federal, state, and local sources when providing such programs. [8] For example, in the early 1980s, the Office of the Inspector General in the Department of Education unsuccessfully requested that seven Texas school districts and the Texas Education Agency refund nearly six million dollars, partly because of the large number of English-proficient students participating in federally-funded BE programs in the state (San Miguel, 1984). [9]

Despite these criticisms, many education policymakers continue to advocate ELA programs without necessarily enforcing the pedagogy or monitoring the programs’ outcomes. If these criticisms reflect reality, language minority students in ELA programs would be expected to perform below their English-immersed peers. To provide insight into this issue, I now analyze whether these programs in practice affect the English-skill acquisition and academic progress of language minority students in U.S. schools.

NELS:88 Data

The empirical work in this study utilizes the 1988, 1990, and 1992 panel data from the Restricted-Use National Education Longitudinal Study (NELS:88) provided by the National Center for Educational Statistics (1996). In 1988, the National Center for Education Statistics (NCES) sponsored NELS:88 to nationally represent eighth graders in both public and private schools, and to track these students over time.

Advantages with Using NELS:88

One advantage of NELS:88 over other educational data sets to test the questions posed here pertains to its extensive language minority questionnaire, which allows ELA participation to be identified before 1988 and between 1988 and 1992. NELS:88 also includes information to observe English-skill acquisition over time, and further provides data on students’ households and schools. Because many LEP students drop out prior to high school, other national datasets (such as High School and Beyond) originating at the secondary school level may unintentionally create a censoring bias in empirical work. Finally, NELS:88 contains cognitive exam scores designed by the Educational Testing Service (ETS), allowing for cross-sectional and longitudinal analyses of academic achievement.

Disadvantages with Using NELS:88

At the same time, the reader should be aware of some disadvantages with NELS:88. The main limitation pertains to the fact that some students selected for the base-year survey did not participate because of a mental or physical disability, or a language barrier. This may lead to an undercoverage bias of up to five percent of the total NELS:88 sample (Owings et al., 1994; in studies analyzing language minority students or recent immigrants, the undercoverage bias is most likely higher. As a result, the reader should keep in mind that the following results pertain to those students who had at least minimal English skills in the eighth grade. [10] Smaller drawbacks with using NELS:88 include the lack of information on students’ English proficiency, academic performance, and type of ELA received prior to the eighth grade. Despite these potential problems, NELS:88 remains a rich dataset to analyze the questions posed here at a national level.

NELS:88 Sample Selection

Of the 16,489 students who participated in the 1988-1992 NELS:88 longitudinal wave, I analyze the 1,560 individuals who spoke a non-English language at home in 1988, resided in a household where a non-English language was usually spoken, reported a non-English native language in 1992, and had information on household socioeconomic status (SES) in 1988. Technical Appendix A (available from the author) lists the specific sample selection process. Native English speakers and those from monolingual-English households axe excluded from the analyses because presumably they would not qualify for ELA, and also because consistent unmeasurable differences between these students and language minorities most likely exist.

Defining ELA Participation

High school ELA participation is gathered from the questions on whether the language minority student received special help in the reading, writing, or speaking of English since the beginning of the ninth grade or since the fall of 1989. When this information is missing, ELA participation may be supplemented from questions asking if the student had been in a bilingual or bicultural course/program, or in an ESL program during high school. [11] Determining ELA received in primary school stems from information on whether the individual had been enrolled in an ELA program for students with a non-English native language.

The specific ELA program can be determined from questions on the form of the special help, or if the student received ELA and reported a bilingual or bicultural course/program or ESL in high school. It should be noted that some students reported taking both BE and ESL in high school. If neither program is reported, I define the ELA program as “other ELA”. Unfortunately, distinguishing between the programs leads to two plausible problems. First, the potential non-response bias on the questions related to the form of the special help exceeds NCES standards; this problem lessens because information can be supplemented with data on high school enrollment in bilingual/bicultural courses or ESL. Moreover, while the 1992 dropout survey asks about the type of special help received, the 1990 and 1992 dropout surveys do not provide data on high school coursework (e.g., bilingual/bicultural courses or ESL). When the 1992 school-dropout survey has missing information on ELA, some dropouts could be assigned to “other ELA ” even if they had taken BE or ESL. To mitigate these concerns, I conduct the following investigations twice: first using overall ELA participation, and second, distinguishing between the different types of programs.

Construction of the English Proficiency Measure

One measure of English fluency comes from how well the individual reported speaking, reading, writing, and understanding English in the 1988 and 1992 surveys. While at first glance one may be skeptical about self-reported data, similar English-skill questions in the 1980 and 1990 U.S. decennial censuses have been used in a multitude of empirical studies, as those in Note 4. Moreover, the demographic statistics presented below provide reliability for these categories because they are consistent with the profiles obtained when analyzing similar English-skill categories in the 1980 and 1990 censuses.

In 1988, the possible English-skill categories for students in language minority households consist of “very well”, “pretty well”, “well”, and “not very well”. In 1992, the available responses for those with a non-English native language include “very well”, “well”, “not well”, and “not at all” Because of the change in coding between the two surveys, I combine the two lowest categories in 1992 because “not at all” was not an option in 1988; presumably, students in this category would have responded “not very well” in the 1988 survey. Taking the minimum response from the questions on the speaking, reading, writing, and understanding of English, I construct a composite measure of English fluency for 1988 and 1992. [12] The 1992 English measure contains three values, with “0” representing the lowest, and “2” the highest, level of English fluency. This information allows English skills to be observed before and after receiving ELA in high school. In many studies, only the ex-poste fluency or achievement can be o bserved, leading researchers to speculate the initial skills in an attempt to obtain unbiased empirical results.

Selected Sample Characteristics

Table 1 presents characteristics of the language minority sample, and the appendix describes the construction of additional variables. NELS:88 includes weights to be used in estimation procedures to maintain a national representation of the base-year and follow-up samples; all analyses utilize the 1988-1992 panel weights in the 1992 survey. Table 1 also breaks down the sample according to whether ELA was received between 1988-1992, as well as by the 1988 English-skill categories. These sample partitions should unveil specific traits unobserved when inspecting the average language minority student.

Table 1 reveals that nearly 21% of the entire language minority sample received some type of ELA in high school. Moreover, the share of students in ELA after 1988 monotonically decreases with the 1988 English-skill levels as expected, given that poor English proficiency influences placement into ELA programs. Also, the English-fluent who went through such programs in elementary school tended to do so in earlier grades than the less fluent. These patterns signal the reliability of the English-skill categories because they reflect conventional wisdom. The proportions of students in BE and ESL appear similar, except for the least English-fluent students who were slightly more likely to receive ESL. Because some students reported taking both BE and ESL, the shares of the three types of ELA do not sum to one.

Table 1 also indicates that the majority of the students report being fluent in English; out of the entire sample, only four percent claim the lowest proficiency level (“not very well”) in 1988. Recall from above that the NELS:88 1988-92 panel ignores some individuals who could not complete the 1988 NELS:88 survey because of mental/physical handicaps or severe language deficiencies (see Note 10). Table 1 reaffirms that this sample mostly contains students who had at least a minimal degree of English fluency in 1988. Despite reporting being English proficient in the eighth grade, many students received ELA in high school.

Other characteristics in Table I are standard. Namely: (a) students receiving ELA tended to reside in lower socioeconomic status (SES) households; (b) household SES rose with English proficiency; (c) ELA students lived in neighborhoods with higher Hispanic shares–a trait pronounced in the less proficient categories; (d) the less fluent tended to report more dissatisfaction with the teaching; (e) Hispanics and Native Americans were more likely to receive ELA, and to report lower English fluency than other racial/ethnic groups; (f) the immigrant sample shares decrease with English skills; (g) both the level and growth in Item Response Theory (IRT) exam scores of ELA students mirror those in the least fluent category; and (h) a relatively higher segment of ELA students dropped out of school.

The observed differences between individuals receiving ELA and their peers suggest that ELA participation is appropriately measured. In addition, the patterns for individuals reporting different fluency levels provide credibility to the English-skill composites; had the sample statistics appeared uncorrelated with the fluency composites, their reliability to gauge “true” English skills would be questionable.

Some of the demographics in Table 1 raise questions which echo some of the above criticisms. For example, why would students who reported being proficient in English in the eighth grade receive ELA in high school? Similarly, why would so many U.S. natives be exposed to high school ELA programs after eight years of U.S. education? While one cannot precisely answer such questions given the data, empirical analyses may be used to gain insight into a related question: what impact do ELA programs have on the English-skill acquisition and scholastic progress of language minority students?

Empirical Framework and Results

I now explore whether ELA schooling programs affect the English-fluency acquisition, academic progress, and school dropout behavior of language minority students following the eighth grade. If ELA affects any one of these aspects, such programs as currently implemented could influence recipients’ future socioeconomic opportunities. To conserve space, the following tables only provide the results on the variables of interest (i.e., ELA programs), although additional results may be obtained from the author.

ELA and English-Skill Acquisition

To analyze the relationship between ELA programs and English-skill acquisition, I estimate an ordered logit model with the 1992 English-skill composite as the dependent variable. The ordered logit technique is suitable when the dependent variable hosts categories that can be ranked in a systematic pattern. In this case, I rank English ability as whole numbers from 0 to 2, where a value of 0 denotes the lowest English fluency level, and 2 represents the highest. Only individuals with a non-missing value for the 1992 English-skill composite are included in this analysis. Specifically, I estimate:

(1) [EP.sub.92] = f(ELA.sub.88-92], [ELA.sub.pre-88], [EP.sub.88], Other, [Location.sub.88]),

where [EP.sub.88] and [EP.sub.92] represent English proficiency in 1988 and 1992; [ELA.sub.88-92] designates whether the individual received ELA between 1988 and 1992; and [ELA.sub.pre-88] denotes prior initial ELA exposure in grades 1-3 or 4-8. The vector Other includes variables for household SES, race/ethnicity, gender, immigration status, immigrants’ time in the United States, private school status, students’ dissatisfaction with teaching, and 1988 IRT-exam composite quartiles; and [Location.sub.88] contains the percentage of Hispanics and other minorities in the 1988 school’s zipcode, and whether the 1988 school existed in an urban, suburban, or rural area. These variables are coded to zero when information is missing, and the regression further includes binary variables indicating missing information.

The inclusion of [EP.sub.88] allows the estimated coefficients to exhibit the “value-added” effects of the right-hand side variables on English fluency, and also controls for the selection process into ELA programs between 1988 and 1992. The use of [EP.sub.88] reduces the extent of the potential bias on [ELA.sup.88-92] because initial English ability itself determines ELA placement; the closer [EP.sub.88] reflects actual English ability (as indicated in Table 1), the better the control for selection between 1988-1992, and the more reliable are the estimated coefficients on [ELA.sub.88-92] (Maddala, 1994, pp. 263-264).

Prior ELA exposure is broken down into grades 1-3 and 4-8 because many LEP students receive ELA in early grades, and because ELA timing may differently affect scholastic progress (e.g., Board of Education of the City of New York, 1994; Lopez, 1998). The remaining variables control for factors that potentially affect English-skill acquisition and scholastic progress. For example, ethnicity, gender, immigrant status, and neighborhood composition relate to English fluency (e.g., Davila & Mora, 2000; Espenshade & Fu, 1997; Lazear, 1999; Stevens, 1992). Unlike the continuous IRT exam scores, the IRT quartiles combine both mathematics and reading scores, which should measure comprehensive academic ability. Dissatisfaction with teaching proxies for the perception of low schooling quality.

Table 2 presents the empirical results from estimating Equation. (1) for the ELA variables. Note that language minority students who received some form of ELA after 1988 reported significantly lower English-skill acquisition on average than their peers, holding other factors constant. That is, even for individuals who reported the same eighth-grade English-skill level, high school ELA programs seemingly related to a relative stagnation or possible deterioration of English skills.

Initial exposure to ELA during grades 1-3 did not significantly affect English-fluency development in high school. The negative and statistically significant coefficient on initial ELA exposure during grades 4-8 may indicate a further deterioration of English fluency associated with relatively late-entry into ELA. Alternatively, the negative results may stem from potential social problems or cultural differences that developed as students entered these programs in later grades. Also, students first exposed to ELA after the third grade may not have had early access to these programs, such as immigrants arriving to the United States in later grades. Although Equation (1) accounts for immigrants’ time in the United States, it is not possible to completely disentangle these possibilities using NELS:88 because individuals are unobserved before the eighth grade.

Table 2 further shows that the type of high school ELA program differently affected the accumulation of English skills. Students who received ESL reported similar development in English fluency as their English-immersed peers, while those in BE or other forms of ELA had lower fluency acquisition. The finding of smaller English-skill development among students who received BE or non-ESL ELA programs between 1988 and 1992 lends empirical support to the criticisms mentioned above. Alternatively, these results may reflect lower self-esteem of ELA students because of a stigma associated with these programs, which in turn affected their reported English fluency. Another explanation may be the conditioning to report lower proficiency because students hear they should be in ELA due to their limited English fluency. If the latter alternatives drove the results in Table 2, however, one would expect students in ESL to also report lower fluency, which does not seem to be the case.

ELA Programs and Academic Progress

In addition to affecting English proficiency, ELA programs may influence scholastic progress. To investigate this premise, an academic progress equation that accounts for household, school, and peer effects can be estimated:

(2) [IRT.sub.92] = a([ELA.sub.88-92], [ELA.sub.pre-88], [IRT.sub.88], [EP.sub.88], Other, [Location.sub.88]),

where [IRT.sub.92] and [IRT.sub.88] denote 1992 and 1988 IRT scores. The remaining terms are defined above, with the exception that the vector of Other excludes the IRT exam quartiles because of the inclusion of the [IRT.sub.88] corresponding to the specific academic discipline. Controlling for [IRT.sub.88] accounts for the fact that initial achievement itself may influence subsequent academic progress, and also reduces the effects of omitted factors such as ability, on achievement (Hanushek, 1986). If [IRT.sub.88] depends on the same ability that affects post-1988 ELA selection, holding [IIRT.sub.88] constant reduces the impact on the estimated coefficient of [ELA.sub.88-92] in Equation (2) (Maddala, 1994, pp. 263-64). That is, the use of [IRT.sub.92] as the dependent variable with [IRT.sub.88] as a regressor controls for selection bias in ELA programs. The capacity to observe the pre- and post-[ELA.sub.88-92] achievement (IRT scores) reflects a major advantage of NELS:88.

I estimate Equation (2) for four different academic disciplines: reading, mathematics, history, and science; Table 3 displays the regression results for the ELA programs for those students who took both exams. I utilize the natural logarithm of the 1988 and 1992 IRT exam scores, such that the coefficients correspond to percentage differences rather than raw differences. The same basic results hold when estimating Equation (2) with the actual IRT scores.

The coefficient on high school ELA is negative and statistically significant in all four regressions, indicating that students in these programs made lower academic progress following the eighth grade than their otherwise similar peers, holding initial ability constant. The largest difference occurred in reading, where students in ELA gained eight percent less on average than their counterparts. The lower cognitive advancement for students in high school ELA programs reinforces the negative ELA effects found above, and may be explained by some of the criticisms associated with the current implementation of bilingual programs.

While students in ELA before the fourth grade did not make statistically different progress in high school, initial exposure to ELA between grades 4-8 negatively related to high school reading and science progress. As with the English-fluency results, initial ELA exposure during grades 4-8 may reflect external, factors associated with late-entry into ELA; it is interesting, however, that late ELA exposure did not have an “across-the-board” effect on academic progress.

When distinguishing between the type of high school ELA programs, all of the ELA coefficients are negative, but not necessarily significant. For example, students who received BE made lower progress in history during high school, but not in the other disciplines, ceteris paribus. Language minority students who took ESL or another form of ELA gained less in reading in particular, and to a smaller extent, in science (and in history for ESL students) than their counterparts in monolingual-English classes. Other ELA programs further related to a relatively large disadvantage (over a 19% difference) in mathematics. These findings indicate some relative academic weaknesses in specific ELA programs. For example, it may be that ESL programs place adequate emphasis on mathematics, but not on reading, history, and science. ELA programs besides BE and ESL seem to sufficiently cover history, but not other subjects. In all cases, students in the specific types of high school ELA programs did not surpass their English-ins tructed peers on average.

ELA Programs and School Attrition

I next investigate the relationship between ELA programs and the decision to drop out of school. To test whether ELA affects student attrition, I estimate the following:

(3) [Dropout.sub.88-92] = d([ELA.sub.88-92], [ELA.sub.pre-88], [EP.sub.88], Other, [Location.sub.88]),

where [Dropout.sub.88-92] represents a binary variable equal to one if the individual dropped out of school between 1988 and 1992, and the remaining terms are the same as in Equation (1). This attrition equation holds constant the “traditional” factors used to explain dropout behavior, such as low SES, poor English skills, immigration status, race/ethnicity, neighborhood or peer effects, and academic motivation (e.g., Ehrenberg & Brewer, 1994; Mora, 1997; Rumberger & Larson, 1998; Velez, 1989).

Table 4 displays the logit results from estimating Equation (3). Individuals who received some form of ELA in high school do not appear significantly more or less likely to drop out of school than their English-immersed peers during this time. Part of this result may be driven by the fact that students who received ELA in high school by definition remained in school at least long enough to take the program. Students first exposed to ELA between the fourth and eighth grades had significantly higher dropout rates than their peers, again possibly reflecting some of the negative effects associated with late-entry into ELA programs. Exposure to ELA before the fourth grade does not appear significantly related to high school attrition.

Separating the types of high school ELA programs again reveals program differences. While language minority students in high school BE and ESL had similar dropout patterns as other language minority students, those who received other forms of ELA were more likely to leave school. Recall, however, that dropouts receiving “other ELA” may have in fact taken BE or ESL in high school because the questions in the dropout surveys do not thoroughly distinguish between these programs; as such, the coefficients on these two programs could be downwardly biased in Table 4.

Concluding Remarks

The empirical results using NELS:88 fail to uncover positive evidence of the effectiveness of high school ELA programs on the English-skill acquisition, academic progress, and educational attainment of language minority students who had attained at least a minimal degree of English fluency in the eighth grade. Specifically, I find that NELS:88 students who received some form of ELA in high school: (1) reported significantly lower English fluency acquisition; (2) made lower academic progress in reading, mathematics, history, and science; and (3) seemed as likely to drop out of school as their English-immersed peers. These findings hold constant “traditional” explanatory factors, such as household SES, initial English skills, race and ethnicity, neighborhood ethnic concentration, immigration status, academic aptitude, and gender. Exposure to ELA in early grades (grades 1-3) did not appear to affect these scholastic outcomes, although initial ELA exposure between grades 4-8 negatively related to post-eighth gra de English-skill acquisition, reading and science development, and staying in school.

It is important to emphasize, however, that the 1988-1992 NELS:88 panel mostly contains students who had at least a fundamental grasp of English in the eighth grade. Recall that the 1988 survey excluded some students because of handicaps or severe language deficiencies (see Note 10). Perhaps high school ELA programs yield different outcomes than those presented here when focussing exclusively on non-English-speaking students, such as recently-arrived immigrants. My results indicate that placing language minority students with a basic knowledge of English into high school ELA programs may be a disservice to the students as well as a misallocation of school resources.

When distinguishing between specific types of high school ELA programs in NELS:88, I find that particular programs differently affected scholastic outcomes. For example, while students who received BE in high school reported significantly lower English-skill acquisition and made smaller improvements in history, they appeared statistically similar to their English-immersed counterparts with respect to school attrition and progress in other academic disciplines. In addition, students in high school ESL programs made relatively smaller advances in reading, history, and science, but had similar mathematics progress and dropout patterns as their peers in monolingual-English classes. ELA programs besides BE and ESL related to lower English-skill acquisition, slower academic growth except in history, and higher attrition rates. Because English proficiency and scholastic progress affect labor market outcomes, extrapolation of these results indicates that particular ELA programs as currently implemented could thwart some language minority students’ opportunities for future economic success.

Explanations for the non-positive results may be found in the growing list of criticisms associated with currently implemented ELA programs, such as segregation into lower quality educational arenas, improperly trained teachers, mismatched language environments, lower levels of academic material, and incorrect placement of English proficient students into ELA programs. Future research on this issue should be conducted using more comprehensive longitudinal data that contain a large sample of non-English speakers, as well as information on the ELA program quality (including teachers’ certification and language proficiency), labor market outcomes, and possible non-income attributes (such as quality of life, socialization issues, or intergenerational effects). Given that education services for LEP students cost between two and three billion dollars per year (American Legislative Exchange Council, 1994), cost-effectiveness should also be considered. If future research finds similar results to those presented here , education policymakers and school administrators may need to better coordinate instructional methods to ensure equal educational and labor market opportunities for this nation’s increasing language minority population.

Marie T. Mora is associate professor in the Department of Economics and International Business at New Mexico State University.

Notes

I would like to express my appreciation to all of the PSJ reviewers (particularly Reviewer #3) and Uday Desai for their insightful comments on this paper. I also thank Alberto Davila, Finis Welch, Manuelita Ureta, Donald Deere, Gail E. Thomas. and Mark Hugo Lopez for helpful suggestions and discussions when I was working on a preliminary version of this study. Given that the Restricted-Use NELS:88 dataset contains individually identifiable information, researchers wishing to replicate the empirical results in this study must contact the National Center for Education Statistics (NCES), U.S. Department of Education. Washington, DC 20208; (202) 219-1920 to obtain a license for the data. The conditions of my license with NCES forbid me from releasing the data myself.

(1.) Proposition 227 is the so-called “Unz initiative” or “English for the Children”. It should be noted that Ronald Unz is not an educator nor an expert on language acquisition or cognitive development. Although the majority of voters passed this proposition, educators and cognitive scientists did not necessarily support this initiative.

(2.) Some exceptions include Danoff (1978). Baker & de Kanter (1981). Lopez and Mora (1998), Lopez (1998), and Greene (1998). August & Hakuta (1997) provide a critical review of the literature.

(3.) Of course, ELA programs may affect other non-academic skills such as socialization. For example, Rodgers (1995) discusses how bilingual education programs may foster friendship and communication between LEP students better than immersion in monolingual-English classes.

(4.) Social scientists have been investigating the link between English proficiency and earnings during the past few decades. Some of the pioneers in this research area are McManus, Gould, and Welch (1983), and Grenier (1984). More recent studies on this issue include Bohara, and Saenz (1993); Chiswick & Miller (1995); Trejo (1997); Mora & Davila (1998); and Davila and Mora (2000).

(5.) Many schools use a combination of screening methods when selecting students for these programs, including teacher observations and referrals, home language surveys, students’ academic records, parent recommendations, oral interviews, language exams, and achievement tests (Fleischman & Hopstock, 1993; Han et al., 1997 U.S. Department of Education, 1992).

(6.) A less common form of BE, known as a “two-way” or dual language program, combines both language minority and language majority students in the same classroom, with the goal that each group learns the other’s language and cultural characteristics (e.g., Griego-Jones, 1994).

(7.) A related issue involves the for ELA programs to serve as a shield for monolingual native-language instruction, sometimes because teachers have limited English ability (Porter, 1996).

(8.) While federal funding has been made possible through the Bilingual Education Act–Title VII of the 1968 Elementary and Secondary Education Act, most BE funds come from state and local sources (U.S. Department of Education, 1992).

(9.) In a recent scenario, a lawsuit filed by a group of parents in Albuquerque, NM, alleged that “to get money for bilingual education, APS singles out English-speaking students for language instruction simply because of their national origin or Hispanic surname” (Franck 1998, p. A1).

(10.) NELS:88 attempted to correct this problem by using a Spanish questionnaire in later surveys and reintroducing the “base year ineligibles”. In 1992, less than half of the 312 base-year ineligibles reported a non-English native language. Unfortunately, NELS:88 does not contain a 1988 survey for these students, such that they cannot be included in analyses utilizing the 1988-1992 panel.

(11.) I do not define ELA participation when an individual reports taking a bilingual or bicultural course (program) or ESL in high school, but claims he or she had not received special help in the reading, writing, or speaking of English. The wording of the bilingual/bicultural questions could classify foreign language courses, ethnography or sociology classes; hence it would be misleading to define these courses as ELA. In fact, of the 4,903 individuals who reported taking a bilingual or bicultural course/program in the 1992 survey (Question F2S13C), 4,263 were native English speakers.

(12.) Most students report the same for all four questions. With missing information, I take the value of the remaining responses. Using the maximum response yields the same basic results as those below.

References

American Legislative Exchange Council. (1994). Bilingual education in the United States 1991-1992. Special supplement to The report card on American education. Washington DC.

August, D., & Hakuta, K. (1997), Improving schooling for language-minority children: A research agenda. Washington, DC: National Research Council.

Baker, K. A., & de Kanter, A. A. (1981). Effectiveness of bilingual education: A review of the literature. Washington, DC: U.S. Department of Education, Office of Planning and Budget, and Evaluation.

Board of Education of the City of New York. (1994, October). Educational progress of students in bilingual and ESL programs: A longitudinal study, 1990-1994. New York, NY: Board of Education.

Chavez, M. A. (1995, September 8). Memorandum (RE: 08945027) to Dr. P. P. Horoschak (Superintendent, Albuquerque Public Schools). Denver, CO: U.S. Dept. of Education (Reg. VIII), Office for Civil Rights.

Chiswick, B. R., & Miller, P. W. (1995). The endogeneity between language and earnings: International analyses. Journal of Labor Economics, 13(2), 246-288.

Crawford, J. (1995). Bilingual education: History, politics, theory and practice, 3rd ed. Los Angeles, CA: Bilingual Education Services.

Curiel, H., Rosenthal, J. A., & Richek, H. G. (1986). Impacts of bilingual education on secondary school grades, attendance, retentions and dropouts. Hispanic Journal of Behavioral Sciences, 8(4), 357-367.

Danoff, M. N. (1978). Evaluation of the impact of ESEA Title VII Spanish/English bilingual education programs. Washington, DC: American Institute for Research.

Davila, A., & Mora, M. T. (2000). The English fluency of recent Hispanic immigrants in 1980 and 1990. Economic Development and Cultural Change, 48(2), 369-389.

Davila, A., Bohara, A.K., & Saenz, R. (1993). Accent penalties and the earnings of Mexican Americans. Social Science Quarterly, 74(4), 902-916.

Donato, R., Menchaca, M., & Valencia, R. R. (1991). Segregation, desegregation, and integration of Chicano students: Problems and prospects. In R. R. Valencia (Ed.), Chicano school failure and success: Research and policy agendas for the 1990s (pp. 27-63). New York, NY: Falmer Press.

Ehrenberg, R. G., & Brewer, D. J. (1994). Do school and teacher characteristics matter? Evidence from high school and beyond. Economics of Education Review, 13(1), 1-17.

Espenshade, T. J., & Fu, H. (1997). An analysis of English-language proficiency among U.S. immigrants. American Sociological Review, 62(2), 288-305.

Fleischman, H. L., & Hopstock, PJ. (1993). Descriptive study of services to limited English proficient students: Summary of findings and conclusions. Washington, DC: U.S. Department of Education.

Franck, M. (1998, March 20). Bilingual bias alleged: APS sending wrong kids to classes, lawsuit says. Albuquerque Journal, pp. A1, A5.

Glenn, C. L. (1997, May). What does the National Research Council study tell us about educating language minority children? Research in English Acquisition and Development Abstracts. Amherst, MA: The Research in English Acquisition and Development Institute.

Greene, J. (1998). A meta-analysis of the effectiveness of bilingual education. Unpublished manuscript, The Tomas Rivera Policy Institute.

Grenier, G. (1984). The effect of language characteristics on the wages of Hispanic-American males. Journal of Human Resources, 19(1), 25-52.

Griego-Jones, T. (1994). Assessing students’ perceptions of biliteracy in two-way bilingual classrooms. Journal of educational issues of language minority students, 13(Spring), 79-93.

Han, M., Baker, D., Rodriguez, C., & Quinn, P. (1997). A profile of policies and practices for limited English proficient students, (SASS 1993-94). Washington, DC: U.S. Government Printing Office.

Hanushek, E. A. (1986). The economics of schooling: Production and efficiency in public schools. Journal of Economic Literature, 24(3), 1141-1177.

Headden, S. (1995, September 25). Tongue-tied in the schools. U.S. News & World Report, pp. 44-46

Ho, K. K. (1985). The paradox of immersion in a second language. NABE Journal, 10(?), 51-61.

Lazear, E. P. (1999). Culture and language. Journal of Political Economy, 107(6), S95-S126.

Lopez, M. H., & Mora, M. T. (1998). The labor market effects of bilingual education among Hispanic workers. Research in English Acquisition and Development Perspectives, 5(2), 33-54.

Lopez, M. H. (1998). Does bilingual education affect educational attainment and labor market outcomes? Unpublished manuscript. University of Maryland.

Maddala, G. S. (1994). Limited-dependent variables and qualitative variables in econometrics. New York, NY: Cambridge University Press.

McManus, W., Gould, W., & Welch, F. (1983). Earnings of Hispanic men: The role of English language proficiency. Journal of Labor Economics, 1(2), 101-130.

Meier, K. J., & Stewart, J. Jr. (1991). The politics of Hispanic education. New York, NY: State University of New York Press.

Mora, M. T., & Davila, A. (1998). Gender, earnings, and the English-skill acquisition of Hispanic workers in the United States. Economic Inquiry, 36(4), 631-644.

Mora, M. T. (1997). Attendance, schooling quality, and the demand for education of Mexican Americans, African Americans, and non-Hispanic whites. Economics of Education Review, 16(4), 407-418.

Murphy, K. M., & Welch, F. (1992). The structure of wages. Quarterly Journal of Economics, 107(1), 285-326.

National Center for Education Statistics. (1996, March). National Education Longitudinal Study: 1988-94 data files and electronic codebook system; Restricted use base year through third follow-up ECB/CD-Rom 1996. Machine-readable data file. Washington, DC: U.S. Department of Education.

Ovando, C. J. (1983). Bilingual/bicultural education: Its legacy and its future. Phi Delta Kappan, 64(8), 564-568.

Owings, J., McMillan, M., Ahmed, S., West, J., Quinn, P., Hausken, E., Lee, R., Ingels, S., Scott, L., Rock, D., & Pollack, J. (1994, September). A guide to using NELS:88 data. Washington, DC: National Center for Education Statistics.

Porter, R. P. (1996). On the state of bilingual education 1990-1995: Forked tongue continued. READ Perspectives, 3(1), 5-61.

Rodgers, M. A. (1995). Language, culture, and society: Bilingual education in the United States. In 0. Thomas (Ed.), Race and ethnicity in America in the twenty-first century: Meeting the challenge (pp. 27-36). Washington, DC: Taylor & Francis International Publishers.

Rossell, C. H., & Baker, K. (1996). The educational effectiveness of bilingual education. Research in the Teaching of English. 30(1), 7-74.

Rossell, C. H., & Ross, J. M. (1986). The social science evidence on bilingual education. Journal of Law and Education, 15(4), 385-419.

Rumberger, R.W., & Larson, K. A. (1998). Student mobility and the increased risk of high school dropout. American Journal of Education, 107(1), 1-35.

San Miguel, G., Jr. (1984). Conflict and controversy in the evolution of bilingual education in the United States–An interpretation. Social Science Quarterly, 65(2), 505-518.

Stevens, G. (1992). The social and demographic context of language use in the United States. American Sociological Review 57(2), 171-185.

Trejo, S. J. (1997). Why do Mexican Americans earn low wages? Journal of Political Economy, 105(6), 1235-1268.

Tucker, G. R. (1980, February). Implications for U.S. bilingual education: Evidence from Canadian research. National Clearinghouse for Bilingual Education, Focus, 2(February), 1-4..

U.S. Department of Education. (1992, June). The condition of bilingual education in the nation: A report to Congress and the President. Washington, DC: U.S. Department of Education.

Velez. W. (1989). High school attrition among Hispanic and non-Hispanic youth. Sociology of Education, 62(2), 119-133.

Willig, A. C. (1985). A meta analysis of selected studies on the effectiveness of bilingual education. Review of Educational Research, 55(3), 269-317.

Selected Sample Means for the

Language Minority Students in NELS:88

ELS in High

School [a] 1988

Variable All Yes No

ELA received in HS 0.207 1.000 —

Bilingual ed. in HS 0.097 0.470 —

ESL received in HS 0.095 0.461 —

Other ELA in HS 0.057 0.276 —

1st ELA grades 1-3 0.178 0.157 0.195

1st ELA grades 4-8 0.090 0.130 0.063

1992 English skills [b]

Not well or not at all 0.040 0.170 0.007

Well 0.277 0.429 0.240

Very well 0.683 0.401 0.753

1988 English skills [b]

Not very well 0.040 0.155 0.009

Well 0.105 0.182 0.083

Pretty well 0.325 0.347 0.306

Very well 0.530 0.316 0.603

1988 Household SES -0.688 -0.918 -0.580

Hisp. share, 88 school [b] 0.335 0.393 0.300

Other min. Share [b] 0.247 0.176 0.273

Dissatisfied w/ teaching 0.134 0.135 0.120

Mexican American 0.433 0.486 0.398

Other Hispanic 0.201 0.211 0.199

Asian/Pacific Islander 0.202 0.138 0.238

Black 0.022 0.001 0.012

Native American 0.038 0.066 0.036

Non-Hispanic White 0.100 0.090 0.115

Female 0.520 0.448 0.546

Born outside U.S. 0.384 0.389 0.392

IRT Exam Scores [b]

1988 Reading 23.421 19.943 24.768

(7.841) (6.072) (8.172)

1992 Reading 29.891 23.800 32.173

(10.134) (8.027) (10.150)

1988 Mathematics 32.067 26.768 34.187

(11.075) (8.812) (11.476)

1992 Mathematics 44.820 37.403 48.013

(14.282) (11.885) (14.302)

1998 History 27.540 25.113 28.301

(4.648) (3.928) (4.571)

1992 History 33.416 30.086 34.562

(5.485) (4.329) (5.516)

1988 Science 16.752 15.010 17.391

(4.298) (3.365) (4.512)

English-Skill Category [a]

Well, not Pretty Very

Variable Very Well Well Well

ELA received in HS 0.482 0.220 0.123

Bilingual ed. in HS 0.202 0.108 0.062

ESL received in HS 0.297 0.087 0.044

Other ELA in HS 0.143 0.064 0.029

1st ELA grades 1-3 0.183 0.247 0.135

1st ELA grades 4-8 0.226 0.090 0.054

1992 English skills [b]

Not well or not at all 0.180 0.043 0.004

Well 0.572 0.444 0.101

Very well 0.249 0.513 0.895

1988 English skills [b]

Not very well 0.275 — —

Well 0.725 — —

Pretty well — 1.000 —

Very well — — 1.000

1988 Household SES -1.157 -0.793 -0.487

Hisp. share, 88 school [b] 0.414 0.413 0.266

Other min. Share [b] 0.220 0.210 0.277

Dissatisfied w/ teaching 0.140 0.146 0.124

Mexican American 0.553 0.482 0.370

Other Hispanic 0.114 0.200 0.225

Asian/Pacific Islander 0.184 0.197 0.212

Black 0.055 0.008 0.033/

Native American 0.075 0.059 0.016

Non-Hispanic White 0.061 0.054 0.141

Female 0.486 0.455 0.573

Born outside U.S. 0.443 0.409 0.351

IRT Exam Scores [b]

1988 Reading 19.198 20.798 26.223

(6.181) (6.420) (7.988)

1992 Reading 22.612 27.516 32.831

(7.911) (8.917) (10.1l5)

1988 Mathematics 26.354 29.873 34.914

(8.385) (9.643) (11.655)

1992 Mathematics 37.073 43.029 47.567

(12.260) (13.963) (14.105)

1998 History 24.030 26.569 29.105

(3.477) (3.982) (4.643)

1992 History 30.040 32.092 34.905

(4.672) (4.980) (5.397)

1988 Science 14.425 15.822 17.970

(3.186) (3.965) (4.427)

1992 Science 21.244 18.820 22.405 18.351

(4.298) (4.740) (6.083) (5.210)

School Dropout 1992 [b] 0.169 0.243 0.133 0.332

Number of Obs: 1,560 318 1,037 213

1992 Science 20.423 22.352

(5.370) (6.095)

School Dropout 1992 [b] 0.242 0.078

Number of Obs: 435 898

Source: The 1988-92 panel of the Restricted-Use NELS:88. See text for sample restrictions.

Notes: Standard deviations for exam scores are given in parentheses. The sample sizes are un weighted, although the statistics have been obtained using the 1988-92 panel weights.

(a.) Sample sizes for the columns do not sum to the total sample because of missing information.

(b.) These sample means exclude individuals with missing information.

Ordered Logit Results for the English-Skill Acquisition of Language Minorities Between 1988 and 1992 (Dependent Variable: 1992 English Proficiency Composite)

Combining HS ELA

Programs

Variable Coefficient Standard Error

Any ELA in high school -1.057 [***] 0.200

BE in high school — —

ESL in high school — —

Other ELA in high school — —

First ELA exposure in -0.250 0.192

grades 1-3

First ELA exposure in -0.580 [**] 0.242

grades 4-8

Pseudo R2: 0.267

Number of observations 1,284

Separating HS ELA

Programs

Variable Coefficient Standard Error

Any ELA in high school — —

BE in high school -0.967 [***] 0.307

ESL in high school -0.215 0.324

Other ELA in high school -1.314 [***] 0.279

First ELA exposure in -0.230 0.192

grades 1-3

First ELA exposure in -0.571 [**] 0.244

grades 4-8

Pseudo R2: 0.269

Number of observations 1,284

Source: The 1988-92 panel from the Restricted-Use NELS:88. See text for sample restrictions.

Notes: These results were obtained using the 1988-92 panel weights provided by NELS:88. Other variables not shown (available from the author) include those in the Other and [Location.sub.88] vectors (see text), as well as binary variables indicating missing data on ELA between 1988-92, prior ELA exposure, the 1988 IRT test quartile, 1988 English fluency, race, dissatisfaction with teaching, the 1988 school and immigration status.

(a.) (***.) (**.). (*.) Indicates statistical significance at the 1%, 5%, 10% levels.

Regression Results for the Academic Progress of Language Minority Students Between 1988-92 (Dependent Variables: Natural Logarithm of 1922 IRT Exam Scores)

Variables Reading Mathematics

Combining high school ELA programs

Any high school ELA -0.081 [***] -0.064 [***]

(0.029) (0.023)

First ELA exposure -0.008 -0.004

grades 1-3 (0.023) (0.020)

First ELA exposure -0.081 [**] -0.012

grades 4-8 (0.041) (0.029)

[R.sup.2] .559 .688

Number of observations 1,141 1,141

Separating high school ELA programs

BE in high school -0.001 -0.024

(0.037) (0.028)

ESL in high school -0.105 [***] -0.035

(0.038) (0.032)

Other ELA in high school -0.048 [***] -0.194 [***]

(0.048) (0.041)

First ELA exposure -0.005 0.010

grades 1-3 (0.023) (0.020)

First ELA exposure -0.072 [*] -0.009

grades 4-8 (0.042) (0.030)

[R.sup.2] .562 .693

Number of observations 1,141 1,141

Variables History Science

Combining high school ELA programs

Any high school ELA -0.046 [***] -0.048 [**]

(0.012) (0.019)

First ELA exposure -0.003 -0.003

grades 1-3 (0.014) (0.023)

First ELA exposure 0.009 -0.062 [*]

grades 4-8 (0.026) (0.037)

[R.sup.2]



Comments are closed.