Videogame-based phonetic training for the perception of American English vowels by native Greek listeners

<ns5:p>This study examined the effectiveness of a videogame-based phonetic training program for improving Greek listeners’ perception of two American English vowel contrasts (/ɪ/−/iː/and/æ/−/ʌ/). The game implemented a high-variability phonetic training paradigm that exposed listeners to multiple voices and lexical items, with speech stimuli generated using synthetic voices. Training consisted of approximately 1.5 hours of gameplay conducted in both quiet and multitalker babble, at progressively lower target-to-masker ratios. Before and after training, identification accuracy was assessed in quiet and in babble. The experimental group showed clear performance gains, whereas a no-training control group showed no improvement. Learning was observed for both vowel contrasts, under both quiet and babble conditions. Partial generalization was observed to untrained voices, but not to untrained lexical items. The results suggest that brief game-based phonetic training with synthetic speech can effectively induce phonetic learning and limited generalization across both quiet and adverse listening conditions. Broader and more robust transfer may require more extended training and greater variability in training materials.</ns5:p>