"It is not surprising, then, that a reasonably large number of ELL students may be receiving inappropriate sets of
accommodations when they take their statewide academic assessments," the researchers write. Inappropriate accommodations may hamper student performance in a standardized math tests just as much as having no accommodations at all, the study found.

ELLs differ from one another in their levels of academic achievement, in their early experiences of testing and schooling and in how proficient they are in English and in their first languages, the researchers explain.

"While teachers or specialists may not be able to clearly differentiate recommended accommodations, their "fall back" approach often seems to be to assign all possible accommodations."

In their study of 272 3rd-grade and 4thgrade Spanish-speaking ELLs in South Carolina who were randomly assigned to test accommodations, researchers found that students who received appropriate accommodations performed better in standardized math assessments than students who had no testing accommodations or who were assigned an inappropriate accommodation.

To determine appropriate test accommodations for ELLs before testing, the researchers used a system developed in South Carolina to serve large minority ELL populations. STELLA (the selection taxonomy for English language learners accommodations) collects information on several student characteristics. The information is used with decision-making algorithms to help assign accommodations.

The student characteristics collected by STELLA include:

• a student's familiarity with standardized assessments as administered in US schools,
• English language proficiency levels;
• first language proficiency levels;
• cultural proximity of past schooling experiences to current schooling experiences,
• length of time in US schools;
• consistency of schooling;
• and experience with specific accommodations.

Researchers collected the information through questionnaires from the students' 11 teachers and from the results of a 30 multiplechoice test taken by the students.

In this study, the eight test-accommodation conditions were:

• no test accommodations;
• a picture dictionary;
• a bilingual glossary;
• oral reading of test items in English;
• both oral reading and picture dictionary;
• both oral reading and bilingual glossary;
• both picture dictionary and bilingual glossary or oral reading;
• and bilingual glossary and picture dictionary.

After students were randomly assigned an accommodation, researchers placed the students in three groups for analysis--no accommodation, recommended accommodations and incomplete or not recommended accommodations.

One reason the research on test accommodations for ELLs has been mixed and confusing, the researchers of this study say, is because it has often been conducted with packages of test accommodations that are not tailored to students' individual needs.

"Overall, the study findings suggest that a more methodical approach toward assigning accommodations to ELL students would be beneficial compared to current practices," the Matching ELLs with right test accommodations researchers conclude.

Additional research using systematized decision-making processes is needed to reconsider the effectiveness of many of the accommodations that other work has suggested might or might not be useful for this population, the authors write.

"Do Proper Accommodation Assignments Make a Difference? Examining the Impact of Improved Decision Making on Scores for English Language Learners," by Rebecca Kopriva, Jessica Emick et al., Educational Measurement: Issues and Practice, Fall 2007, Volume 26, Number 3, pp. 11-20.

Published in ERN November 2007 Volume 20, Number 8