Naturalistic mishearings, ‘slips of the ear’, and mondegreens provide data about lexical retrieval, as they instantiate cases where instead of the intended word, a listener accesses an incorrect — but similar word. Data of this sort are rich in terms of their contributions to the language sciences, and have been explored in our prior work to examine phenomena such as segmental similarity and its relation to perceptual confusion matrices. In the present work, we examine the relationship between the token frequency of the intended word and the actually-perceived word, where in fact there is no general trend towards hearing/replacing the intended word with a blanket more frequent word. Instead, however, there is the curious (though consistent) finding that listeners tend to replace the intended word with a word from a similar frequency class (e.g. replacing a low frequency word with a low frequency word). Such results seem at first blush paradoxical — in order to replace the misheard word with a word of a similar frequency, you’d need to have heard and processed the frequency of the word you didn’t hear. In fact, we will propose that listeners estimate the frequency class of the word they misheard based on durational and sentence-level inferences, thereby demonstrating “graceful degradation”, whereby the misheard word doesn’t exhibit totally wild guessing. Data from both conversational and song-lyric databases will be analyzed and discussed with respect to models of lexical access, with the broader goal of demonstrating the contribution of error data to psycholinguistic models of perception.
A graceful degradation account of lexical retrieval – evidence from naturalistic misperception
Andrew Nevins (UCLA) and Kevin Tang (Yale)
Monday, November 2, 2015 - 4:00pm to 5:00pm
109 Grove StreetNew Haven, CT 06510