High school teachers’ perceptions of using ChatGpt for designing English reading comprehension tests for tenthgrade students

Abstract

Reading tests are effective assessments to evaluate EFL learners’ reading ability; however, designing them has been reported by teachers to be time-consuming and demanding. With the development of AI tools like ChatGPT, many English teachers have employed them to assist in making such tests. While ChatGPT can support the generation of many reading test components, some concern that its created content is not reliable without human supervision. To investigate high school teachers’ perceptions of using ChatGPT for this purpose, we carried out this study using embedded mixed-methods research. We distributed questionnaires to 20 high school teachers who were generally familiar with the tool; then, to gain an in-depth understanding of their opinions, we conducted semi-structured interviews with 10 randomly selected participants. After analyzing the data, we found that between 60% and 95% of the teachers agreed ChatGPT is useful in generating reading tests across various aspects, particularly in generating comprehension questions (95%) and saving time and effort (85%). However, 75% of respondents also recognized that certain issues persist in ChatGPT’s output, such as repetitive or mechanical phrasing in questions, inconsistencies in passage length and structure, and potential inaccuracies or biases. This highlights the crucial role of teachers in reviewing and refining AI-generated content to ensure its suitability before using it in assessments.