Not all errors are the same: ERP sensitivity to error typicality in foreign accented speech perception

Publication date: Available online 22 March 2018
Source:Cortex
Author(s): Sendy Caffarra, Clara D. Martin
Intercultural communication has become more and more frequent in the recent globalized society. When native listeners try to understand non-native speakers, they have to deal with different types of grammatical errors, some being frequently encountered and others being less common. The present ERP study investigated how native listeners process different types of morphosyntactic errors in foreign accented speech and whether they are sensitive to error typicality. Spanish natives listened to Spanish sentences in native and foreign (English) accent. ERPs were recorded in response to morphosyntactic violations that were commonly (gender errors) encountered in English accented Spanish or not (number errors). Although sentence comprehension accuracy did not differ across accents, the ERP responses changed as a function of accent and error type. In line with previous studies, gender and number violations in native accented speech elicited LAN-P600 responses. When speech was uttered by foreign speakers, number violations (uncommon errors) showed a P600 effect, while gender violations (common errors) did not elicit late repair processes (reflected by the P600) but an N400 effect. The present results provide evidence that the neural time course of parsing depends not only on speaker's accent, but also on input error typicality.



https://ift.tt/2HWsRQk