The proposed research seeks to extend our knowledge about how early and in what ways children can produce elements capable of being perceived as a string of a given language. I will explore an innovative way of assessing speech intelligibility in young bilinguals and monolinguals by comparing the proportion of utterances correctly identified by listeners for each of the two groups in a context-free situation. Listeners will be blind to target, language, and condition of the child (bilingual or monolingual). The tapes to be analyzed contain vocalizations produced by 31 children, 11 English/Spanish bilinguals, 13 English monolinguals and 7 Spanish monolinguals during play sessions recorded in our lab between the ages of 16 and 30 months. A digitizing utility will be used that allows one to extract selected utterances from tapes and record them into the computer. By using this procedure, any parental or child speech preceding or following the selected utterance is removed so that the listeners are left with no contextual information to bias their judgments. The analysis will thus compare the percentages of utterances from each of the two groups that listeners are able to associate with the right target and will further examine the nature of the cues used for language identification.
