Comparison of Psychometric Properties of Turkish as a Foreign Language Listening Tests Composed of Independent and Common-Stem Items

This study comparatively examined the psychometric properties of Turkish listening tests composed of independent and common-stem multiple-choice items that measure the same construct in the context of teaching Turkish as a foreign language. Additionally, students’ perceptions of these two different item formats were evaluated. The research was designed as a descriptive study using a relational survey model. The study group consisted of 201 international students enrolled at the B1 level in Turkish Language Teaching Centers during the 2024–2025 academic year. The assessment tools were developed on the Concerto platform, and the tests were administered online in a single session across three stages. According to the findings, comparisons of item difficulty indices revealed statistically significant differences and moderate to large effect sizes in some item pairs, indicating that item format may influence student performance. Similarly, item discrimination indices showed significant differences in several item pairs, although both tests overall consisted of highly discriminative items. Analysis of test mean scores showed that the independent item test yielded significantly higher performance at the 0.05 level. The reliability coefficients of both tests were high, with no statistically significant difference between them. Based on students’ perceptions, the independent item test was found to be more advantageous in terms of time management and online usability. In contrast, the common-stem item format was perceived as more cognitively demanding. Both item types received similarly positive feedback regarding item clarity and ease of answering listening questions. These results suggest that in teaching Turkish as a foreign language, independent items may contribute to higher student achievement and improved item discrimination in listening assessments.