Abstract
This paper reports on a project aimed at improving the pronunciation of a group of Japanese EFL high school students, none of whom the author has met in person, through collective feedback videos. Conducted in collaboration with a public high school in Eastern Japan, the project involved all 1st-year students recording oral readings of their opinions on school uniforms, created through a guided composition format. The author listened to about half of these recordings and produced five 5- to 10-minute feedback videos. Students watched the videos in class and were encouraged to review them outside of class as well. They then re-recorded their oral readings of the same scripts and completed online surveys on their experience. For segmental features, the quality of four consonant phonemes (/l/, /r/, /θ/, and /ð/) distributed across 26 different words in the pre- and postintervention recordings was analyzed. Prosodic features were assessed holistically by comparing the first 15 seconds of the pre- and post-intervention readings. Results indicated that overall pronunciation of the targeted phonemes significantly improved, though with a small effect size that varied by phoneme. Prosody also improved, with a much larger effect size. Survey results were in line with these findings, with students reporting that they learned the most from the two videos focused on prosody.
https://ror.org/05cw8ax51 
