States Scramble to Rewrite Language-Proficiency Exams
States across the country are scrambling to craft better tests for students with limited English skills in response to stringent new timelines imposed by the "No Child Left Behind" Act of 2001.
As early as this week, the federal government was expected to award grants of up to $2 million each to states working to improve their testing systems. At least three consortia of states are applying for a chunk of the $17 million available under the "enhanced-assessment grants" competition to design better tests for English- language learners.
In addition, California and New York are preparing to unveil new language-development tests of their own, while Massachusetts officials hoped to announce a request for proposals late last month.
The No Child Left Behind Act, the latest reauthorization of the Elementary and Secondary Education Act, requires states and school districts to test the English proficiency of all English-language learners beginning this school year. States, districts, and schools that receive money for language instruction under Title III of the law must show annual increases in the number and percent of students who become proficient in English, as well as in the number who make progress toward that goal.
Title I of the law also requires states, districts, and schools to make yearly progress in the proportion of limited-English-proficient students who score at a proficient level on state tests of academic content.
"The difficult piece has been the speed with which they want you to do this: today," said Jeff Nellhaus, the state assessment director in Massachusetts.
New Tests Needed
A number of language-proficiency tests are currently on the market, such as the Language Assessment Scales, published by CTB/McGraw-Hill, and the Woodcock- Munoz Language Survey, published by Riverside Publishing/Houghton- Mifflin.
But "most of the instruments that are out there were not designed to evaluate student progress," said Wesley D. Bruce, the director of school assessment for Indiana. "They were more placement and diagnostic instruments."
Moreover, experts say, many of the tests are cumbersome or time-consuming to administer. They don't provide teachers with enough information to improve instruction. And they were not designed to reflect states' academic standards. Those tests also may not measure all five domains that the federal law requires: comprehension, reading, writing, speaking, and listening.
"They're dated," said Charlene Rivera, the executive director of the Center for Equity and Excellence in Education, based at George Washington University. "There really hasn't been much development over the last decade or more."
California, an early leader among the states, has been working with CTB/McGraw-Hill to write the California English Language Development Test since 1997. The state is revising the exam to make it easier to administer and hopes to have it in schools by June.
The initial version was administered to children one-on-one. "But once we started using it statewide, then there were problems," said Jeanette B. Spencer, an education consultant to the California education department. "A one-on-one test for 1.5 million students," she said of the estimated number of English-language learners in the state, "is asking a lot."
New York state has been working with the Princeton, N.J.-based Educational Testing Service to devise a series of language-development tests for students in grades K-12. The ETS, which has contributed at least $1 million toward the effort, plans to make the package of tests, which can be customized, available to other states and districts.
Mari A. Pearlman, the vice president of the company's division of teaching and learning, said the ETS has been working with teachers to make the tests immediately helpful to them. "We actually asked teachers what they thought it was relevant and important to know about language proficiency in kids," she said.
On the writing portion, for example, the scoring guide is designed to give teachers a picture of a student's strengths and weaknesses as they score the exam, "so at a glance, you can get some kind of instructionally useful information," Ms. Pearlman said.
The ETS, which has spent more than $5 million overall on the development of language-proficiency instruments over the past few years and expects to unveil the exams in early 2004, is also designing professional development for teachers linked to the assessments, said Ms. Pearlman.
What's more, the company is working with one of the consortia applying for federal grants. Staffed by AccountabilityWorks, a nonprofit policy research and consulting organization based in Washington, the consortium includes Florida, Maryland, Michigan, Pennsylvania, and Tennessee.
"We think that there's a need for new and better tests that are standards-based and that are consistent with the best research," said Theodor Rebarber, the president of AccountabilityWorks. "Part of our goal is to have these assessments be useful not only for accountability, but also for supporting instruction."
The largest consortium, helped by the Washington-based Council of Chief State School Officers, includes 16 states and the American Institutes for Research, a nonprofit behavioral and social-science research group also based in Washington, as the test-maker.
"This is probably the most intense thing I've ever seen," said John Olson, the director of assessments for the council's State Collaborative on Assessment and Student Standards. Leaders of the LEP- SCASS, as the project is known, have organized the effort. He said they hope to have an initial test available by May.
Paul M. La Marca, the director of assessments for Nevada, the lead state in the consortium, said the goals of such tests have shifted toward gauging how students perform relative to state academic-content standards. "It's more for instruction and accountability," he explained, "not simply for the classification of whether a student is limited-English-proficient or not."
Added Ms. Rivera of George Washington University: "The question always is, at what point does a person have enough English to be able to appropriately and validly take a content-area test," or participate in an English-only class in math or science.
"It's been a question that a lot of people have been trying to understand better," she said, "and no one has been able to really come to the point where they say they have the magic answer."
In fact, research suggests a difference between the everyday- language proficiency required in social settings and the more complex academic language needed to succeed in school.
As an example, Ms. Pearlman points to the expression "greater than" in mathematics.
"If you're just learning English, nobody uses the word 'great' anymore as a size word," she explained. "So you can imagine how confusing this can be."
According to Frances A. Butler, a senior research associate with the Center for the Study of Evaluation, located at the University of California, Los Angeles, existing language-proficiency tests tend to focus more on the social or conversational uses of language and less on academic language.
"Our research has indicated that, to some extent, there's a mismatch between the content on the existing language-proficiency tests and the language that students have to know on the large-scale content assessments that they take every year," she said.
"The existing language tests simply don't go far enough in providing the information we need," said Ms. Butler, "to know whether students can handle the material that they need to be able to handle to do well in school. It's been a problem for years. It's just that it's getting attention now because of No Child Left Behind."
Together with colleague Alison L. Bailey at UCLA's Center for Research on Evaluation, Standards, and Student Testing, Ms. Butler is trying to formulate a framework for the types of evidence that test- developers should gather to help design the next generation of language- proficiency tests. Those include the language teachers use with students, the variety of textbook materials, the language demands on tests of academic content, and the language in state and national content standards.
Authentic Measures Missing
Measured Progress, a testing company in Dover, N.H., is working with a group of Western states to pursue that idea of academic-language proficiency.
The company is working on a set of English-language-development standards and a test design. "Our goal is to do so in the context of how students use English as a speaking, listening, reading, writing, and comprehension skill within various academic settings," said Edward Roeber, the vice president of new business for Measured Progress.
"The shortcomings of some of the current measures are [in] their failure to really put kids in authentic situations in a science class, in a social studies class, in an English class, in a math class," he said. "Because they are more generic measures of English, they fail to predict success in those academic settings."
Scholars, however, warn that separating the language proficiency needed to do well in specific content areas from knowledge of the content itself will be difficult.
Lyle F. Bachman, a professor in the department of applied linguistics at UCLA, suggests that, in the end, English proficiency and content knowledge may be inextricably linked, because language is essentially a tool used to process information about specific content.
The more tasks are designed to isolate English proficiency, he suggests, the more they look like nonlanguage. The more they are designed to measure higher-order aspects of language use, the more they include topical knowledge and skills.
Meanwhile, test-makers and states are forging ahead. Harcourt Educational Measurement plans to set proficiency scores for its new Stanford-English Language Proficiency Test next month. The tests, which will be available in the spring for grades K-2, 3-5, 6-8, and 9- 12, measure all the language skills required under the ESEA using one instrument, as opposed to separate tests for listening, speaking, and writing, said Diane Johnson, a senior assessment specialist for Harcourt.
"It's a very complete test," said Margarita Miska, a senior director for catalog development. "We have done a lot of research on what teachers will need, based on the new legislation."
"There's no doubt that major publishers will fill the void," said Mr. Bruce of Indiana, a member of the CCSSO consortium. "But if you have one more potential test to choose from, the likelihood that you'll get one that is the best match for your state is improved."
Ms. Rivera of George Washington University warned that states have an even tougher task ahead.
"The attention right now is on this idea of how do you assess language proficiency," she noted, "and they really need to put as much attention, if not more, on how to assess content for students who are English- language learners. They need to place attention on how to include students appropriately in [state content] assessments."
Vol. 22, Issue 14, Page 10