1-153 - Retention of statistical learning

Session Type: Paper Symposium

Integrative Statement

Infants show a remarkable ability to learn highly frequent information in language. However, for practical reasons, most work tests infants immediately after training. Here we investigate how much infants retain after exposure to highly frequent information. Paper 1 assesses retention of words segmented from an artificial language after 80-minute and 24-hour delays, showing poor retention at 6.5 months of age but better retention at the 80-minute delay in infants with greater amounts of slow-wave sleep. Paper 2 investigates how long it takes infants to learn a novel label they can then use in the service of segmenting words with iambic stress at an age when infants typically favor segmentation of words with trochaic but not iambic stress patterns. Infants required many days of exposure to use a novel label to segment iambic words the next day. Paper 3 investigates infants’ memory for recently segmented words in older 22-month-olds. Words with high internal predictability supported novel label-object mapping after a 10-minute delay, whereas words with low internal predictability did not. However, when infants were given the opportunity to map both word types immediately after familiarization, the meaning representations were successfully retained following a delay. A discussant will synthesize the findings and their implications for understanding language acquisition. Together these findings suggest that while memory for newly acquired information remains fragile early in development, many of the types of experiences infants have in their daily lives support retention.

Sub Unit

• Language, Communication

Chair

• Rebecca Gomez, The University of Arizona

Discussant

• Jenny R. Saffran, University of Wisconsin–Madison

Individual Presentations

• Fragile Long-Term Retention Following Statistical Learning in Infancy - Presenting Author: Lucia Sweeney, University of Arizona; Non-Presenting Author: Katharine Simon, The University of Arizona; Non-Presenting Author: Rebecca Gomez, The University of Arizona

• Segmentation is easy. Long-term retention in the service of bootstrapping is hard. - Non-Presenting Author: Michelle Sandoval, Wenatchee Valley College; Presenting Author: Rebecca Gomez, The University of Arizona

• Statistical learning supports word learning and memory - Non-Presenting Author: Ferhat Karaman, University of Tennessee; Presenting Author: Jessica Hay, University of Tennessee, Knoxville