Early in 2005, I attended a conference for newspaper education reporters in Texas on the effects of No Child Left Behind (NCLB). One of the stars of the event was Sandy Kress, a former education advisor to President George W. Bush and one of the bill's chief architects (Pyle, 2005). Kress gave a powerful and passionate defense of NCLB, describing how, for the first time in history, the federal government was serious about educating all students, how it was finally doing something about closing the achievement gap and improving education for poor and minority students. With the air of the attorney he is, he rejected the criticisms of NCLB made by the speakers who preceded him. However, when asked about English language learners, he chuckled and then admitted that no one really knew what to do with them. All he knew, he said, was that they needed to be included somehow in testing and accountability programs.
Included...Somehow
This lack of clarity regarding English language learners (ELLs) is most obvious in the way that NCLB handles the Limited English Proficient (LEP) subgroup. As with other subgroups, such as African Americans or Latinos, the LEP subgroup is expected to make adequate yearly progress (AYP) toward proficiency. By 2014, all English language learners, regardless of how long they have been in the United States, must pass their state's accountability tests. Moreover, if the requisite number of English language learners in a school's LEP subgroup does not pass the tests in a given year, the school is deemed as failing and may be subjected to sanctions.
When it comes to English language learners, NCLB defies logic. Common sense dictates that if you administer a test to students in a language they don't understand, they probably won't do well on it. NCLB contradicts itself on this very point. The law describes a limited English proficient student as one “whose difficulties in speaking, reading, writing, or understanding the English language may be sufficient to deny the individual the ability to meet the State's proficient level of achievement on State assessments” (Sec. 9101(37)). Nevertheless, NCLB mandates that English language learners do just that, and schools are punished if they don't.
The biggest flaw, however, is the way AYP is calculated for the LEP subgroup. The LEP subgroup is treated the same as other subgroups, such as those based on ethnicity/race. But unlike students in the ethnicity/race subgroups, who obviously remain in the same category, students continually move in and out of the LEP subgroup. Even more problematic, those students who speak the most English—and thus who are more likely to pass the test—leave the LEP subgroup only to be replaced by newly arrived English language learners who speak the least English. This makes it impossible for the LEP subgroup to show consistent growth.
In partial acknowledgement of these flaws, then–U.S. Secretary of Education Rod Paige announced in February 2004 two modifications that states could take advantage of: They could exclude LEP students from the reading/language arts test during their first year in the United States, and they could include English language learners who had attained English proficiency in the LEP subgroup for up to two years.
These changes are insufficient, however. First, few English language learners can learn enough English in one year to pass their state's reading test. Second, the students must still take their state's math test their first year in the United States, even if they arrive on the day of the test. And contrary to what some might think, math tests do not pose less of a challenge to English language learners than English tests do. These tests include a great deal of language, which might explain why English language learners in Arizona scored lower in their state's math assessment than they did in reading and writing (Wright & Pu, 2005). As for keeping former English language learners in the LEP subgroup for two additional years, this strategy only delays the problem rather than solves it.
“Reasonable Accommodations”
NCLB requires schools to test English language learners in a “valid and reliable manner.” Schools are to provide “reasonable accommodations,” including, “to the extent practicable,” administering tests in students' native languages until the students attain proficiency in English. Although these requirements sound reasonable, they actually introduce several problems. First, the law neither defines “accommodations” nor defines what constitutes “reasonable.” Second, no monitoring mechanisms are in place to ensure that schools actually provide accommodations. Third, there is no consensus on what constitutes an acceptable accommodation; thus state accommodation policies vary substantially (Rivera & Collum, 2006). Finally, little research has been conducted on the most effective accommodations for English language learners, and the research that does exist on how schools can provide accommodations while maintaining test validity and reliability is inconclusive (Abedi, Lord, Hofstetter, & Baker, 2000). In short, NCLB requires something that we don't yet know how to do.
“Accommodation” in Arizona
A survey that I conducted of teachers in 40 Arizona elementary schools with substantial English language learner populations revealed that fewer than half of these schools provided accommodations on the state test (Wright & Choi, 2006). Because teachers received conflicting information about accommodations, practices varied widely. Some teachers were allowed to read the directions or questions aloud in English; others were allowed to translate them. Some teachers permitted students to use English or bilingual dictionaries. Most teachers reported that these accommodations were of little benefit, however. One teacher noted that her school had only five Spanish-English dictionaries to share across 27 classrooms. Even when students had a dictionary, few actually used them. Many didn't know how; others were embarrassed to use them in front of their peers. Even in those instances in which translation was permitted, few students requested it.
Teachers from the Arizona survey described the behaviors of their English language learners as the students took the state test: Seventy-eight percent reported instances of students complaining that they could not understand the test questions, 68 percent reported instances of students leaving entire sections of the test blank, and 78 percent reported instances of students randomly bubbling in answers without reading the questions. Although these teachers believed that schools should be held accountable for their effectiveness in teaching English language learners, 78 percent thought that high-stakes tests were not appropriate for this purpose, and 90 percent thought that these tests did not provide an accurate measure of English language learners' academic achievement.
Teachers also worried about the effect of these tests on students. During testing, 88 percent of teachers surveyed had observed English language learners becoming visibly upset, 71 percent had seen students cry, 68 percent had seen students get physically ill, and 35 percent had seen students throw up.
Testing Math in Texas
Texas has extensive “linguistically accommodated testing” (LAT) guidelines for its state math test. For a study of a Texas intermediate school, I served as a volunteer mentor/tutor for two newcomer 5th grade students from Cambodia—Nitha and Bora (Wright & Li, in press). Neither spoke English when they arrived two months into the school year. Nevertheless, they were required to take—and expected to pass—the state math test in English. Despite the extensive LAT guidelines, Nitha and Bora took the same math test as their English-fluent peers, and with no accommodations. Bora got seven of the 44 questions correct; Nitha, who was more skilled in math, only got six correct. Their teacher reported (with a laugh) that during the test, Bora bubbled in five answers in a row and then enthusiastically yelled out “Bingo!” How's that for testing English language learners in a “valid and reliable manner”?
Before the test, I provided Nitha and Bora with weekly math tutoring sessions in Khmer (Cambodian). The school hoped that I could provide translation assistance on the day of the test. However, the state denied this request because I was not a district employee. Then the school hoped that I could provide a glossary of math terms in English with Khmer translations. The state also denied this request because the classroom teacher could not verify that the glossary only contained word-for-word translations rather than explanations.
LAT procedures indicated that Nitha and Bora could ask their teacher to read test questions aloud. Even though the students understood that they could request this, they never did. Despite their progress learning English in the six months that they had been in the United States, they simply didn't understand enough of the language to comprehend math questions read aloud.
Even if they had been English-proficient, they still lacked sufficient math knowledge to answer the questions. Nitha and Bora came from an impoverished village in Cambodia. Their school lacked running water and electricity, and their teachers had less than a high school education. There were not enough textbooks for the students, and the math curriculum was substantially below U.S. standards. Despite being good students in Cambodia, Nitha and Bora were far behind their U.S. peers.
The Problems with Native-Language Tests
Given the myriad of problems associated with testing English language learners in English, testing students in their native language appears to be the best accommodation. However, native-language tests are only required “to the extent practicable”—and few states have found them practicable. Translating a state test into another language doesn't automatically produce a valid and reliable instrument covering the same concepts at the same level of difficulty. Creating valid and reliable native-language tests requires following the same procedures that produce a valid and reliable English language test. This is an expensive and time-consuming process that few states can afford to undertake. Although the majority of English language learners in the United States are Spanish speakers, the remaining English language learners represent more than 400 different language groups. The numbers of Spanish-speaking English language learners may be high enough to justify creating tests in Spanish—that approach would certainly make sense for students in Spanish/English bilingual programs—but how many states would find it practicable to develop tests in Chinese, Hmong, Arabic, Urdu, and Af Maay?
In addition, non-English tests only work when English language learners can read, write, and receive content instruction through their native language. Developing tests in Khmer, Lao, or Vietnamese is not feasible because few Southeast Asian American students can read or write in their native languages. Even for students like Nitha and Bora, who can read Khmer well, a Khmer version of the math test would have been of little help because nearly all their math instruction has been in English since their arrival in the United States.
Manipulating the Numbers
Given these technical flaws and the empty promise of accommodations, many states now understand the best strategy for LEP subgroup success: It'snot having an LEP subgroup. Some states are setting a high cutoff for the number of students that a school must count as a subgroup. For example, in Arizona, schools need at least 30 English language learners at the same tested grade level to have an LEP subgroup (Arizona Department of Education, 2003). In 2004, this minimum group size rule enabled 680 Arizona schools to avoid a failing designation (Kossan, 2004). Other states, such as Texas, have set their minimum group size at 50 students or more. This has helped many Texas schools avoid having an LEP subgroup, and not a single school in the state has had an Asian or Native American subgroup (Castro, 2005).
Each state can determine its minimum subgroup size, and thus far the federal government has made no effort to standardize how states determine that number. NCLB is vague on this point. The only schools that end up having LEP subgroups are those with substantial English language learner populations, and these schools are generally overcrowded, underfunded, and have the least experienced teachers.
Until recently, states were also able to negotiate separate deals with the U.S. Department of Education. For example, during the past few years, Arizona had approval to exclude from AYP calculations the test scores of English language learners who were enrolled for fewer than four years (Kossan, 2004). This strategy alone excluded many English language learners, particularly those who were least likely to pass the tests. In July 2006, however, the Department of Education informed Arizona—and other states that had these exclusions in effect, such as New York, that it will no longer allow these exclusions. As a result, more than 122 additional schools in Arizona will be labeled as failing to make AYP (Kossan, 2006). Arizona's superintendent of schools has filed a lawsuit against the federal government in an attempt to maintain these exclusions of English language learner test scores (Moravcik, 2006).
Apparently, few teachers and school administrators realize that exclusions of their English language learners' test scores are taking place. In the Arizona study, 98 percent of the teachers reported being under pressure to raise English language learners' test scores. In addition, 93 percent of the teachers said that preparing students for high-stakes tests was preventing them from focusing on the students' linguistic and cultural needs. Few of their schools required English as a second language (ESL) instruction, and most did not even have ESL curricular materials. Rather than teach ESL, teachers were required to provide a test-preparation–based curriculum. Even when teachers were able to raise the scores of their English language learners, they did not equate this with effective teaching.
Only English Spoken Here
An additional problem with NCLB is its exclusive focus on English. Although bilingual education is still permitted (if state governments allow it), the term “bilingual” has been stripped from federal education law along with any recognition of the individual and societal benefits of bilingualism (Wright, 2005). Even after-school heritage-language programs are being affected because low-scoring English language learners are often required to attend after-school test-preparation programs instead.
Federal education policy is out of touch with the United States' need for bilingual citizens. It's also out of touch with the priorities and goals of other federal agencies. Foreign language experts testified at the National Briefing on Language and National Security in 2002 that “the U.S. government's language capabilities remain grossly inadequate” and that “we need more linguists in more languages at higher levels of proficiency than ever before.” A job recruiter from the U.S. State Department expressed her frustration with finding heritage language speakers with sufficient proficiency in their native languages to qualify as translators. She lamented, “The attempt made in our private and public schools to annihilate any knowledge of the language spoken at home has been very successful” (National Foreign Language Center and National Security Education Program, 2002). Very successful indeed. In this crucial area of need, NCLB is a giant leap backward.
It's Just Common Sense
Teachers are not opposed to accountability, but most agree that English language learners should be excluded from the regular state tests, at least until they are proficient enough in English to meaningfully participate. Moreover, given NCLB's internal contradictions—such as requiring students who don't speak English to pass tests in English and setting expectations that are mathematically impossible to attain—it's no wonder the U.S. Department of Education allows states to use a variety of strategies to ensure that the fewest number of schools end up with LEP subgroups. Because so many states are already excluding the scores of so many English language learners from their AYP calculations, why make students take the test at all?
Most of the problems of testing English language learners go away if we simply have patience and give students the opportunity and support to first learn English and then learn academic content. Let's encourage schools to adopt high-quality language programs that ensure that English language learners become the well-educated bilingual citizens that the United States so desperately needs. And let's remove the burden of preparing English language learners for state tests from teachers so they can focus instead on their students' linguistic, cultural, and academic needs.
