Learning non-native tone through artificial grammar: The role of prosodic background

Previous research suggests that a speaker’s first language (L1) shapes how non-native phonological contrasts are perceived and learned. Tone-languagespeakers often outperform non-tone speakers in simple tone discrimination tasks, but this advantage can diminish in more complex learning contexts (e.g., Hwang et al., 2021; Perdomo & Kaan, 2021). This study investigates how prosodic differences within a single language family affect the learnability of tonal patterns. Using an Artificial Grammar Learning paradigm, we compare two Korean dialects: Kyeongsang Korean (KK; N = 30), a pitch-accent language, and Seoul Korean (SK; N = 22), an intonation language. Participants were trained to identify plural forms in an artificial language, where tone distinguished singular and plural. They were assigned to either a high-tone or low-tone learning group and tested on novel items. Preliminary results show that SK participants performed similarly across tone conditions, while KK participants learned significantly better in the low-tone group (p < 0.001). Interestingly, SK participants outperformed KK participants in the high-tone group, whereas the opposite was true in the low-tone group. Accuracy was highest when the tone of the singular form matched that of the plural reduplicant. These findings highlight how prosodic background shapes tone learning in subtle but systematic ways.