Effects of energetic and informational masking on speech segmentation by native and non-native speakers

In this study, we asked whether native and non-native speakers of English use a similar balance of lexical knowledge and acoustic cues, e.g., juncture-specific allophones, to segment spoken English, and whether the two groups are equally affected by energetic masking (a competing talker) and by cognitive load (a simultaneous visual search task). In intact speech, as well as in both adverse conditions, non-native speakers gave relatively less weight to lexical plausibility than to acoustic cues. Under energetic masking, overall segmentation accuracy decreased, but this decrease was of comparable magnitude in native and non-natives speakers. Under cognitive load, native speakers relied relatively more on lexical plausibility than on acoustic cues. This lexical drift was not observed in the non-native group. These results indicate that non-native speakers pay less attention to lexical information-and relatively more attention to acoustic detail-than previously thought. They also suggest that the penetrability of the speech system by cognitive factors depends on listener's proficiency with the language, and especially their level of lexical-semantic knowledge. (C) 2010 Elsevier B.V. All rights reserved.