Using DeepL translation tool to address speaking anxiety

Abstract

This study examines how Japanese EFL students may utilise DeepL, one of the online translation tools (OTTs), to help them overcome speaking anxiety to increase natural discourse. In this study, 31 third-year students at a Japanese public junior high school were given a conversation task with DeepL in the same pair at the beginning of seven consecutive lessons over three weeks. Before and after the tasks, pre- and post-assessments were conducted and video-recorded. A mixed approach was employed to analyse the collected data. This mixed method study examines the process of DeepL use, interactions, and speaking output by video-recording at the pre- and post-assessment and using questionnaires after the post-assessment. The results showed an increase in the number of T-units (a sentence with more than one subject and verb) and the length of T-units (words per T-unit) in pairs. However, translation errors occurred while utilising DeepL. Therefore, students need to enhance their language and computer skills in everyday lessons. On the other hand, with DeepL, students with speaking anxiety actively participate in paired speaking activities. To clarify the discrepancy, therefore, the students' unrecognised speaking anxiety was identified by analysing questionnaires and T-tests. The students who used DeepL every time in the tasks tended to correlate with physical changes in speaking anxiety.