Effects of frequency and neighborhood density on spoken-word recognition in noise: Evidence from spoken-word identification in Dutch

Hintz, F. 1 & Scharenborg, O. 1, 2

1 Centre for Language Studies, Radboud University, Nijmegen, The Netherlands
2 Donders Institute for Brain, Cognition, and Behavior, Radboud University, Nijmegen, The Netherlands

Theories of spoken-word recognition assume that frequent words and words with sparse phonological neighborhoods are more often recognized correctly than infrequent words and words with dense phonological neighborhoods, respectively. While the effects of frequency have been found to be robust across numerous studies, neighborhood density effects appear to be more fragile and even absent at times. In the present study, we tested the involvement of these item properties in spoken-word recognition in adverse listening conditions. Dutch participants took part in a word identification experiment, using CVC words presented in the clear and in various noise degradations (SNRs: 6, 2, -2, -6). The words varied orthogonally in frequency and neighborhood density. Recognition rate decreased linearly with deteriorating listening conditions. Replicating earlier studies, we found frequency to influence recognition performance consistently in all noise conditions. The effects of neighborhood density were more subtle but statistically robust. In contrast to our predictions, the analyses revealed a recognition advantage for high density words in noise. Complementary analyses additionally showed the importance of the frequency of the word?s phonological neighbors for successful word recognition. Specifically, words whose neighbors had a low rather than a high mean frequency were more often recognized correctly.