APPLIED LINGUISTICS REVIEW, vol.13, no.4, pp.461-475, 2022 (AHCI)
The use of translation for language teaching and assessment, by and large, has been abandoned with the adoption of audio-lingual and communicative approaches in language teaching. As a result, nowadays translation items are not commonly used for measuring language proficiency in international language proficiency tests (e. g. TOEFL, IELTS). However, there are several countries that still use translation items in their national language proficiency tests (e. g. Turkey, Japan, China, Romania among others). The present study aims to examine whether or not multiple-choice translation items are an appropriate tool for measuring proficiency in English. To this end, the perceived level of difficulty and validity of multiple-choice translation items in the National English Proficiency Test (YDS) in Turkey were examined. The findings revealed that the participants did significantly better on the translation items than on the rest of the test items. They also perceived the translation items as the easiest among all the rest items in YDS. Moreover, while YDS as a whole indicated a strong validity based on correlation with TOEFL PBT Reading Sample Test, the translation items indicated moderate validity. Importantly, there was a significant difference between the two correlations. These findings suggest that multiple-choice translation items are likely to lower the overall validity of YDS tests, inflate the scores of test-takers and, thus, might be considered as problematic for the quality of the tests.