Bergelson, E. & Swingley, D.
University of Pennsylvania
Most theories of word recognition presuppose that listeners interpret speech largely in terms of language-specific consonant and vowel categories that can be identified independently of the talker. In contrast, many accounts of language development argue that infants' lexical representations are holistic, and do not support the same sorts of generalizations as adults?. Yet, few studies address the degree to which infants? interpretation of words in speech relies on lexically relevant phonetic variation as opposed to lexically irrelevant variation such as talker identity (e.g. Singh, 2008). Here we present two studies that build on recent work showing that 6-12 month olds fixate pictures of foods and body parts named by their mother (Bergelson and Swingley 2012). Study 1 examined whether infants show speaker-invariance in interpreting word meanings. An experimenter rather than the infant's mother named pictures. Infants (6-12 months; n=58) performed above chance, looking more to the named picture, both in the 6-8 month subset and overall (all p<.05 by Wilcoxon Test), showing no hindrance in understanding the novel talker. Study 2 examines whether infants are sensitive to single-phoneme changes when comprehending known words. Here, mothers labeled words for their infants, but each word was mispronounced, such that a single vowel was changed, e.g. ?hund? for ?hand?, ?opal? for ?apple?). 12-14 month-old infants show significantly worse performance in this study than an age-matched group from the original study (p<.05). Testing with 6-8 month olds is in progress. Together, these studies suggest that before age one, infants? lexical comprehension seems to be appropriately flexible; that is, young infants? ability to understand word meanings does not rely on specific voice characteristics of their most familiar interlocutors, but deteriorates when these interlocutors change a single phoneme in the noun to which they are drawing their infants? attention.