This article describes a transcription task cycle that was designed to facilitate the development of skills for the IELTS (International English Language Testing System) speaking test at a language school in Japan. The cycle involved practice test, transcription, student correction, teacher correction, and retrial of the original test and continued over a number of weeks. This investigation aimed to evaluate the nature of any linguistic gains made and the likelihood that items corrected by the student or teacher would appear in the test retrial and later weeks. The results showed that students' approaches to correction varied widely from single corrections to complete reformulations, and that students were able to make gains with a variety of forms, appearing more likely to use their own corrections than those of the teacher. However, the short timescale and number of students limit generalization as to the stability of those gains.