The influence of neighborhood density on the recognition of Spanish-accented words.

Abstract

Foreign-accented speech is more difficult to recognize than the same words produced by a native speaker because the accented speech may activate many additional competitors, or it may strongly activate a single, but incorrect, word during lexical retrieval. Experiments 1 and 2 examined the recognition of native-produced and foreign-accented words varying in neighborhood density with auditory lexical decision and perceptual identification tasks, respectively. Experiment 1 found increased reaction times (RTs), especially for accented dense words. Analysis of misperceptions from Experiment 2 found that the mean number of phonologically distinct misperception tokens was higher for native than accented stimuli, suggesting that accented speech does not tend to activate more lexical candidates. Furthermore, a higher proportion of misperceptions in the accented condition (71%) compared with the native condition (58%) was accounted for by the most frequently reported misperception token, suggesting that accented speech instead tends to strongly activate 1 particular neighbor of the target word during lexical competition. Moreover, systematic phonemic substitutions in the misperceptions suggest that lawful acoustic-phonetic variations introduced by the accented speaker's L1 (native language) play a crucial role in determining which neighbor is activated as a strong competitor.