The Validity of Translation Items in English Language Proficiency Exams in Turkey

Abstract

While using translation items for measuring language proficiency is not common in language testing in the world (e.g. TOEFL, IELTS) such items are commonly used in Turkey. In the two standardized language proficiency exams, YDS and YÖKDİL, translation items are given in a multiple-choice format. Using them as a tool for measuring language proficiency raises a question about the validity of the items. Validity in language testing refers to the suitability of an item to the purpose of the test, and traditionally, it can be measured via correlation of the test score with a valid test. In this respect, the validity of the translation items in YDS and YÖKDİL was investigated together with the overall validity of these exams via correlating the scores of 32 participants with those of TOEFL PBT Reading Sample Test. Afterward, t-tests were conducted to see the compatibility of the difficulty level of translation items with the overall exams to see if the participants found them easier. Laslty, a translation task consisting of 20 sentences taken from the published e-YDS items by ÖSYM were given to the participants to find out the relation between the participants’ production and recognition skills of translation. It has been found that YDS and YÖKDİL are both valid (.88 and .82 respectively). However, the translation items lower their validity coefficient; and especially for YDS, translation items group has the lowest correlation coefficient (.61). With the t-tests, it has been found the participants did significantly better on the translation items in both of the tests. Also, the production translation task indicated moderate validity (.66) and it was found to be more difficult than YDS, YÖKDİL and the translation items in these tests. It can be concluded that multiple-choice translation items lower the quality of the English language proficiency exams in Turkey.