Aiming for advanced intelligibility and proficiency using mobile ASR

This experimental study aimed to determine the impact of mobile-based Automatic Speech Recognition (ASR) in Gmail on intelligibility and proficiency, as well as whether any individual factors influenced learning outcomes. It focused on 26 Intermediate learners of French as a foreign language enrolled in two university courses geared towards the development of advanced oral skills but with different approaches to integrated instruction. It innovatively combined human-based and machine-based ratings within an ecological paradigm, following Levis’s (2005) intelligibility principle and Thomson and Derwing’s (2015) call for research that is readily useful for language instructors. Results show that ASR users significantly outperformed non-ASR users on intelligibility, particularly when exposed to instruction on spelling-to-sound patterns, and demonstrated the biggest growth in proficiency. Gender was also found to impact results. Pedagogical implications and venues for future research are offered.