2-041 - Specificity of Infant Statistical Learning

Session Type: Paper Symposium

Integrative Statement

Over the past three decades, research has revealed that infants have remarkable computational abilities that allow them to segment speech by tracking statistical regularities in their input. However, little is known about the specificity with which infants represent newly extracted units. Across three papers, this symposium will explore how English-learning infants represent fine-grained details about syllable stress and co-occurrence statistics in recently segmented words. Results from Paper 1 suggest that 8-month-old infants recognize recently segmented words even when the stress pattern between familiarization and test is altered from a trochaic to iambic stress pattern (i.e., CAsa í caSA). Paper 2 shows that 13-month-old infants do not simply chunk syllable sequences together when they extract them from speech, but also represent syllable order information. Finally, using a segmentation + word learning task, the third paper reveals that 17-month-olds will only map an extracted label to a novel object if the exact syllable sequences are preserved; they fail to map labels to objects when syllable co-occurrence patterns are violated. Integrating across these findings, the discussant will suggest that early in development, infants may have relatively weak or crude representations of newly segmented words that become more precise across the second year of life.

Sub Unit

• Language, Communication

Chair

• Jessica Hay, University of Tennessee, Knoxville

Discussant

• Casey Lew-Williams, Princeton University

Individual Presentations

• How Accurately Do Infants Represent Lexical Stress Information in Recently Segmented Words? - Presenting Author: Sara Parvanezadeh Esfanani, University of Tennessee, Knoxville; Non-Presenting Author: Jessica Hay, University of Tennessee, Knoxville

• Extraction of Syllable Order During Word Segmentation - Presenting Author: Lucia Sweeney, University of Arizona; Non-Presenting Author: Rebecca Gómez, University of Arizona

• Specificity of Statistically Segmented Words - Presenting Author: Katharine Graf Estes, University of California, Davis; Non-Presenting Author: Dylan Antovich, UC-Davis