Prosodically-Driven Spoken Word Recognition
Elif Canseza Kaplan · 2025
Understanding speech in multi-talker environments is a central challenge in everyday communication, and listeners vary widely in how well they cope. Musical expertise has been proposed as one factor shaping this variability, since music and speech rely on overlapping auditory and cognitive processes. This PhD project examined whether musical expertise and individual differences in musical abilities are associated with spoken language processing, particularly in speech-on-speech conditions. The focus was on how musical background relates to the uptake of prosody during spoken word recognition, considering both local segmental (proximal) and contextual suprasegmental (distal) cues. Across three studies, sentence recall tasks were combined with eye-tracking (Visual World Paradigm) and pupillometry. Study 1 compared musicians and non-musicians: musicians recalled more sentences under speech-on-speech masking, with gaze and pupil measures suggesting differences in attention allocation and lexical competition. Study 2 treated musical ability as a continuous construct (rhythmic, melodic, general musical sophistication) and found that higher rhythmic and melodic abilities were linked to greater use of syllable-duration cues, especially when speech was masked by competing talkers. Study 3 tested contextual prosody, showing that preceding rhythm and intonation biased lexical-stress expectations; both groups used these cues, but musicians showed distinct competition-resolution patterns under masking. Overall, findings indicate that categorical musician–non-musician contrasts are not uniform, and that individual abilities offer a more informative view of real-time prosodic processing. Results are correlational, limited to English and Dutch, and generalisation to other prosodic systems or multilingual contexts requires caution. Future work should adopt longitudinal designs, broader language samples, and ecologically valid paradigms.