Transfer in speech motor learning: The role of voicing

Abstract

Previous studies have demonstrated that American English speakers can improve their production of phonotactically illegal onset clusters (e.g., DBEEGO) after structured practice. However, the nature of what is learned remains incompletely understood. We use a transfer paradigm to address this question by examining performance on trained and untrained novel consonant sequences. In particular, we investigated whether the differences between voiced and voiceless stop-stop clusters (e.g., /gd/ vs. /kt/) influences transfer of learning, hypothesizing that the voiced clusters involve more complex motor control. Forty native speakers of American English practiced nonwords beginning with either voiced (/gd/, /db/, /gb/) or voiceless (/kt/, /kp/, /tp/) stop-stop onset clusters. All participants were tested on both types of clusters at baseline (prior to practice) and in two retention sessions (20 minutes (R1) and 2 days (R2) after practice). Blinded coders rated cluster accuracy based on presence of a vowel in the acoustics. Preliminary results (n = 10) indicate a trend of bi-directional transfer, with participants in both practice conditions exhibiting improved accuracy for both trained and untrained clusters.