proficient enough at a language that they test equitably with native speakers,

according to a recent study in Education Policy Analysis.

"If testing is to serve as an essential tool to inform the improvement of

instructional practices and student learning, how can we have confidence in

using test results that may not provide accurate information about how well a

student is doing at school or in a subject matter?," write the authors, who

analyzed test data for a large population of Chinese and Spanish ELLs in the San

Francisco Unified School District (SFUSD).

"The issue becomes critical when testing results are used for high-stakes

decisions impacting individual students." To address this issue, a study was

conducted from September 2001 until June 2004 to explore the development of ELL

student performance on achievement tests in conjunction with developing English

language proficiency.

Supported in part by a grant from the U.S. Department of Education, the study

tracked performance on the SAT/9 mathematics tests for students from grades 2-5.

When students are enrolled SFUSD, the district collects the student's

demographic information and English language proficiency is assessed. Students

are classified into three categories: English Only (EO), when a student is from

an English speaking background; Initial Fluent English Proficient (IFEP), when a

student is from a non-English background but is proficient in English; and

Limited English Proficient (LEP), when the student is from a non-English

speaking background and is not proficient in English.

The majority of the ELLs at 2nd grade had been in the school district for

three years. To examine language proficiency development, the researchers

tracked correlations between students' performance in math problem solving and

math procedures. Correlations between the two are lower for ELLs than for native

speakers because of language demands of solving math problems, they explain,

i.e. native speakers have smaller differences in scores for math procedures and

solving math problems. The researchers found it took five to six years for

performance of Chinese ELLs and Spanish ELLs in math procedures and problem

solving to converge with the national sample.

"Achievement testing for English Language Learners, ready or not?" by

Sau-Lim Tsang, Anne Katz, and Jim Stack, Education Policy Analysis Archives,

2008 Volume 16 Number 1.