Pilot Program Could Help English-Learners
Computer model to weigh individual testing needs.
As early as next school year, some states may roll out a sophisticated computerized system intended to ensure that English-language learners receive the testing accommodations that fit their individual needs.
As matters stand now, said Rebecca J. Kopriva, the lead developer of the system, teachers can be inconsistent—even within the same school—in how they decide what kinds of accommodations to give children who take part in large-scale standardized assessments.
But the system, devised by researchers at the University of Maryland College Park with $1.7 million in federal funding, aims to help educators identify the most appropriate kind of assistance to offer students with limited English proficiency in the pressured testing arena.
“There’s plenty of evidence that making the wrong decision in giving an accommodation can hurt a child’s test performance,” said Ellen Forte, a consultant in testing and accountability for the Washington-based Council of Chief State School Officers.
She called the newly developed system—the Selection Taxonomy for English Language Learner Accommodations, or STELLA—a “major contribution to the field” and “a huge leap forward from where we have been” in understanding such techniques.
Ms. Forte noted that under the federal No Child Left Behind Act, all states must provide testing accommodations for English-learners, though it’s only optional that they provide some kind of alternative form of the regular content-area tests, such as “plain English” or translated versions.
The testing accommodations can vary widely, from simply providing extra time on a test to offering a bilingual glossary of terms to which students can refer during the exam.
Although some states provide alternative tests for English-language learners, a majority give such students regular tests with accommodations to account for their limited English proficiency. Among the most common accommodations are:
• Bilingual word lists or glossary that a student can refer to during the test
• Administration of the test in a small group
• Reading the test directions or test items aloud to the student in English
• Providing extra time for students taking the test
But Ms. Forte said that many states simply began using the same testing accommodations for English-learners that they used for students with disabilities, though only a small subset of those accommodations were appropriate.
The U.S. Department of Education is investing in STELLA because improving assessment of English-language learners and students with disabilities is a “department priority and a considerable challenge for states in implementing No Child Left Behind,” said Sue Rigney, an education specialist for the student assessment and school accountability program of the department’s office of elementary and secondary education.
“We want better, more accurate assessment results so schools and teachers can make better instructional decisions,” she said.
The system was developed by a team at the Center for the Study of Assessment Validity and Evaluation at the University of Maryland, in cooperation with education officials from Maryland, North Carolina, South Carolina, the District of Columbia, and the 80,900-student Austin Independent School District in Texas. It has been piloted in a number of districts, including two in South Carolina, the 19,300-student Beaufort County schools and the 42,200-student Charleston schools.
In addition to spending nearly $2 million to develop the system, the Education Department has committed $700,000 to refine and implement it.
Ms. Kopriva, who recently stepped down as head of the University of Maryland center, is working with the CCSSO to line up states willing to help refine the system and to try out the test version when it is ready next school year.
South Carolina, the lead state for the project’s development grant, is also the lead state for the current implementation grant. The CCSSO also is a partner in the current grant.
Teri Siskind, the director of assessment for the South Carolina Department of Education, said that the basic system will be free, but that states may have to pay to customize it or maintain it.
The advantage of the computerized system, said Ms. Siskind, is that it takes into account many more factors than teachers typically consider in selecting accommodations.
The system depends on three sources to make a determination on which accommodation is a good fit for the student—a school record, a form filled out by teachers, and a form filled out on the basis of an interview with a parent.
Ms. Kopriva said the researchers have identified 10 factors that are most relevant when deciding what kind of accommodation is appropriate. Educators plug in the data about individual students, and the computer, programmed to make choices based on research findings, does the work to make the match.
For example, she said, the computer will show that a student who is new to the United States and can read well in his or her native language but not in English should be given an alternative test in that native language.
If such an alternative test isn’t available, the student should be given a test that has been adjusted to contain visuals and “plain English,” said Ms. Kopriva.
And if that alternative isn’t available either, she said, the computer will show that the student should receive two accommodations: a bilingual glossary and the opportunity to hear test directions or items read orally in his or her native language.
In focus groups, she said, teachers expressed concern about having enough time to collect the data needed for the system to make the appropriate matches for students.
But Ms. Rigney of the federal Education Department said she believes educators will have an incentive to use the system because research shows they aren’t getting good results on tests with the way they are providing accommodations now.
Vol. 26, Issue 22, Pages 15-16