Adaptation to an unfamiliar dialect of one's native language presents a special case for prosodic learning, since most other aspects of the grammar are held constant. This study explores the representation of two dimensions of tune through a series of experimental tasks in which speakers of American English attempt to directly imitate and then generalize the dialectal features of a native speaker of Glasgow English. The results show that speakers are able to modify both f0 peak timing and f0 excursion in order to approximate the target dialect, and that they do so both during direct imitation and when generalizing to new sentences. The findings suggest that peak timing and excursion are not only represented differently, but that learning progresses differently for the two dimensions in going from direct imitation to generalization.

Index Terms: dialect learning, intonation, peak alignment