Proceedings of the Annual Meeting of the Cognitive Science Society

How does the presence of background noise affect thecognitive processes underlying spoken-word recognition? Andhow do these effects differ in native and non-native languagelisteners? We addressed these questions using artificial neural-network modelling. We trained a deep auto-encoderarchitecture on binary phonological and semanticrepresentations of 121 English and Dutch translationequivalents. We also varied exposure to the two languages togenerate ‘native English’ and ‘non-native English’ trainednetworks. These networks captured key effects in theperformance (accuracy rates and the number of erroneousresponses per word stimulus) of English and Dutch listeners inan offline English spoken-word identification experiment(Scharenborg et al., 2017), which considered clean and noisylistening conditions and three intensities of speech-shapednoise, applied word-initially or word-finally. Our simulationssuggested that the effects of noise on native and non-nativelistening are comparable and can be accounted for within thesame cognitive architecture for spoken-word recognition.