EFFECT OF TRANSLATED VERSION OF TEST AMONG ELL
Mahesh Gurung is the faculty sponsor of this poster.
Abstract
The US is a diverse immigrant nation with increasing population of English language learners (ELL) comprising around 11% of K-12 student with almost a million in 2015. Sultan Turkan and Maria Elena Oliveri discuss on effects of translated versions of tests in consideration for Providing Test Translation Accommodations to English Language Learners on Common Core Standards-Based Assessments. They suggest student performance is likely to relate to the quality of test translation and the approaches used to minimize translation error and maximize equivalence. Based on this article, we reviewed available data from Department of Education website between 2004/2005 to 2015/2016 in math and science where translated version and non-translated of the exams were administered. We randomly selected four states that administered translated version (California, Texas, Oregon and New York) and four that did not (Florida, Illinois, Virginia and Nevada). Our preliminary data indicated performance of non- translated version (mean=34.62) proportionally outperformed translated version (mean=25.84). Since this review was preliminary and does not include all states, results are subject to reservation. Further analysis warranted to falsify our observed patterns.
EFFECT OF TRANSLATED VERSION OF TEST AMONG ELL
The US is a diverse immigrant nation with increasing population of English language learners (ELL) comprising around 11% of K-12 student with almost a million in 2015. Sultan Turkan and Maria Elena Oliveri discuss on effects of translated versions of tests in consideration for Providing Test Translation Accommodations to English Language Learners on Common Core Standards-Based Assessments. They suggest student performance is likely to relate to the quality of test translation and the approaches used to minimize translation error and maximize equivalence. Based on this article, we reviewed available data from Department of Education website between 2004/2005 to 2015/2016 in math and science where translated version and non-translated of the exams were administered. We randomly selected four states that administered translated version (California, Texas, Oregon and New York) and four that did not (Florida, Illinois, Virginia and Nevada). Our preliminary data indicated performance of non- translated version (mean=34.62) proportionally outperformed translated version (mean=25.84). Since this review was preliminary and does not include all states, results are subject to reservation. Further analysis warranted to falsify our observed patterns.