Sequence effects during speech perception reveal multi-accent processing costs

Alternating between different talkers during listening typically incurs a cognitive processing cost. How these processing costs manifest, and potentially differ, in a multi-accent setting remains to be examined. Across two experiments, we investigate (1) whether talker and accent switching costs are driven by engagement of a recalibration mechanism, and (2) whether global listening context affects the magnitude of talker and accent switching costs. The results of our first experiment indicate that switching between speakers of the same second language (L2) accent (e.g., between two Mandarin-accented speakers of English) was less cognitively challenging than switching between speakers of different L2 accents (e.g., between a Mandarin-accented speaker and a Turkish-accented speaker of English). This outcome suggests that the perceptual distance (i.e., the holistic estimate of spectral and temporal differences in acoustic signals) between two speakers' productions determines the size of associated switching costs, such that recalibration is less cognitively demanding for speakers with the same L2 accent. In our second experiment, we examine whether a more challenging block-wide listening context results in a global upregulation of cognitive resources, and, subsequently, reduces the cognitive resources required to (a) process L2 accent and (b) resolve local talker and accent changes. Here, the overall cognitive demands of processing L2 accent were reduced, as predicted, but talker and accent switching costs were not. We conclude that talker and accent switching are supported by a recalibration mechanism and that global upregulation of cognitive resources may reduce L2 accent processing costs but not local switching costs.