Variation in the strength of lexical encoding across dialects.

Lexical processing is slower and less accurate for unfamiliar dialects than familiar dialects. The goal of the current study was to test the hypothesis that dialect differences in lexical processing reflect differences in lexical encoding strength across dialects. Lexical encoding (i.e., updating the cognitive lexical representation to reflect the current token) was distinguished from lexical recognition (i.e., mapping the incoming acoustic signal to the target lexical category) in a series of lexical processing tasks with Midland and Northern American English. The experiments were conducted in the Midland region with Midland and Northern listeners. The results confirmed differential processing of the two dialects: the Midland dialect was processed more quickly than the Northern dialect. The results further revealed significantly larger repetition benefits (i.e., priming) and cross-dialect lexical interference effects for lexical forms in the Midland dialect than in the Northern dialect for both listener groups, particularly when the stimulus materials were presented in noise. These results suggest that lexical information is more strongly encoded for the contextually-local Midland dialect than for the non-local Northern dialect. We interpret these effects as reflecting cognitive processing costs associated with normalization for dialect variation, which lead to weaker lexical encoding under more difficult processing conditions.