The Relationship Between Auditory Working Memory and Statistical Learning in Infancy

Although previous research indicates that there is a strong relationship between auditory working memory and language skills in children and adults, no studies have explored this relationship in infancy. Working memory develops shortly before the onset of many language abilities in infancy, including the ability to use statistical information to segment words from speech. To track frequent syllable co-occurrences within a speech stream to locate word boundaries, infants must hold auditory information in their mind as one syllable quickly displaces another. Thus, it seems likely that auditory working memory plays a role in statistical learning. In the current study, I sought to test the relationship between 8-month-old infants’ auditory working memory capacity and their ability to track regularities to segment words from a naturally produced speech stream. For Experiment 1, I hypothesized that infants would look longer at varying streams than constant streams based on their auditory working memory capacity (2 or 3 items). A 2x2 ANOVA revealed a significant effect of stream type on looking time during the auditory working memory task; follow-up pairwise comparisons revealed that infants looked significantly longer at 2-Varying than 2-Constant streams, but not 3-Varying compared to 3-Constant streams. These findings indicate that 8-month-old infants have an auditory working memory capacity of 2 nonlinguistic items, and they can hold at least 500 ms of auditory information in their working memory. For Experiment 2, I hypothesized that infants would look longer at words with stronger statistical regularities between syllables. A one-way ANOVA revealed that there were no significant main effects of word type on looking time, indicating that infants were not able to discriminate between HTP, LTP, and Novel words during test. Finally, to explore the relationship between infants’ performances in Experiment 1 and Experiment 2, I predicted that infants who looked longer at varying streams during the auditory working memory task would look longer at HTP words in the statistical learning task. However, due to the small sample size, I conducted exploratory correlational analyses, which revealed no significant relationships between infants’ looking times on the auditory working memory and statistical learning tasks.