We present a methodological reflection following an experiment carried out in an interdisciplinary project. Within the framework of this project, we wanted to experimentally test the impact of corrective gestures on the acquisition of segmental and suprasegmental pronunciation skills. A pretest/post-test/control group design was set up with three experimental conditions: a test group that received corrective gestures adapted to the anticipated errors, a test group that received corrective gestures unadapted to the anticipated error, and a control group that received audio-only correction. We recorded 45 Japanese-speaking students of French as a Foreign Language, equally distributed among the three groups. The acoustic and perceptual analyses of the 7200 oral productions (160 *45) did not allow us to differentiate between the three groups. However, the observed limitations raised methodological reflections and also lead us to a new proposal of experimental design.