When is a Dar a Car? Effects of Mispronunciation and Referential Context on Sound-Meaning Mappings

Infants exhibit a remarkable capacity for speech sound discrimination early in life, including sensitivity to the dimensions that define phonemic categories of human languages (e.g. Eimas, 1974; Eimas & Miller, 1980; Eimas, Siqueland, Jusczyk, & Vigorito, 1971; Miller & Eimas, 1983). Over the first year of life, infants’ perceptual sensitivities are refined to reflect the phonological structure of the ambient language (e.g. Kuhl et al., 1992; Werker & Tees, 1984). Such early knowledge of language-specific phonological structure should be beneficial to infants beginning the complicated task of building a lexicon. Word learning requires not only that phonological word forms be stored in memory with appropriate specificity, but also that mappings between these phonological forms and meaning be learned. In the real world this is a daunting task: Every time a word is spoken, the learner has to decide onto which lexical entry it should be mapped or whether the new token is indeed an exemplar of an existing lexical entry. Knowledge about phonological categories at the onset of word learning should enable learners to focus on the linguistic detail critical for making such determinations. Moreover, advance knowledge of important phonological dimensions might free resources, facilitating the task of mapping between phonological form and meaning. However, there has been considerable debate about whether older infants and young children are able to apply this phonological knowledge in tasks that require attention to meaning. Older infants and young children often fail to discriminate even familiar speech sound contrasts in certain word recognition and word learning tasks (for a review, see Barton, 1980). Recently, evidence has accumulated that older infants are able to use phonological information in some referential tasks (Fennell & Werker, 2003; Swingley & Aslin, 2000, 2002; Werker, Corcoran, Fennell & Stager, 2002). Studies using variants of the Preferential Looking Paradigm (Golinkoff, Hirsh-Pasek, Cauley & Gordon, 1987), in particular, have painted a much rosier picture of older infants’ phonological competence during word learning and recognition. There are two advantages to using this type of procedure to assess infants’ phonological