This proposal explores the relationship between language learning and hearing multiple talkers, or talker variability. Perhaps surprisingly, talker variability appears to facilitate learning. Infants, for example, better recognize a word they have heard spoken by multiple talkers. Later in development, children more accurately produce unfamiliar words they have heard spoken by multiple talkers. These findings hint that talker variability might also serve as an intervention tool, helping to facilitate the speech of children with language impairments. The proposal's goal is to flesh out this possibility;applying talker variability's potential to facilitate speech in both children with typical language development (TD) and specific language impairment (SLI), across multiple learning contexts, and using multiple measures of learning. Thus, the anticipated result of the proposal is a highly detailed picture of whether, when, and for whom talker variability facilitates language learning. Talker variability's influence on speech learning is examined across four experiments, each providing a different context in which talker variability might facilitate learning. Two factors, linguistic target and learning modality, are crossed to create the four experiments. The linguistic targets are phonotactics-specific sound sequences within words like the /pt/ in 'helicopter'-and prosodic contours-stress patterns across syllables like trochees such as the noun record and iambs such as the verb record. Both targets are known to be difficult for children with SLI, but only phonotactics have been studied in the context of learning from talker variability. The training modalities are perceptual learning-hearing the training items-and articulatory learning-hearing and saying the training items. The two modalities represent the means through which children might learn from talker variability, but previous research has focused only on perceptual learning. Studying both modalities will also indicate which modality is most relevant to clinical practices. Two other aspects of the proposal are noteworthy. First, learning is assessed in two posttests, so the proposal determines whether talker variability leads to improvements that persist over time. Second, we are measuring learning with a traditional transcription-based measure used in previous research, but also with speech movement data. We record children's lip and jaw movements during their productions of the linguistic targets using an Optotrak camera system. These movement records can then be compared across experimental conditions to assess how talker variability influences a very specific but necessary articulatory component of speech accuracy. Furthermore, the kinematic analyses offer insights into speech accuracy that are not subject to the perceptual biases of transcription. Thus, the proposal will provide a highly objective measure of learning.
Children with specific language impairment have longstanding deficits that affect their speech and language learning. The goal of this proposal is to take research on language learning in typically developing children and translate it into clinically oriented research in children with specific language impairment.