Sinking about speech: Acoustic similarity versus linguistic experience in prelexical processing

Speech sounds of a second language are often hard to pronounce, and speakers approximate the correct pronunciation by using a close relative from their native language. Using an identity Mismatch Negativity (iMMN) design, we examined whether processing of such mispronounced segments is driven by acoustic similarity with the standard pronunciation or by one’s experience. Specifically, we compared the English standard pronunciation of the interdental fricative in the pseudoword ‘thond’ to deviant pronunciations ‘tond’ and ‘sond’, typical of either German (who frequently substitute ‘th’ with /s/), or Dutch second-language learners (who frequently substitute ‘th’ with /t/). Acoustically, /s/ is always more similar to ‘th’ than /t/. ERP-data from Dutch and German listeners were analyzed subtracting the responses to the exact same stimuli presented as deviant and standard across conditions. In Dutch and German participants, both substitutions for ‘thond’ elicited a significant iMMN and consecutive P2. For Dutch listeners, the effects of ‘sond’ and ‘tond’ were equally large in both the iMMN and P2. For Germans, ‘sond’ elicited a smaller P2, while the iMMNs were of the same size. The iMMN results suggest that acoustic distance influences prelexical processing of ‘th’-substitutions in non-native listeners, whereas linguistic experience may have an impact downstream (P2)