Quantity Meets Quality: Examining input statistics for early word learning from the infant’s point of view

Quantity Meets Quality: Examining input statistics for early word learning from the infant's point of view Project Summary / Abstract Children acquire language in everyday contexts where the language environment plays a critical role in supporting early language development. In early word learning, infants learn object names by linking heard words to seen objects in their environment. This is well recognized as a computationally complex problem due to referential uncertainty -- at the moment of hearing any word, there may be many potential referents in the learner's immediate environment. Many laboratory experiments are designed based on various assumptions about the uncertainty challenge that young learners face, which may not reflect the uncertainty encountered in everyday learning experiences. Despite considerable research efforts, little is known about the referential uncertainty infants face in their everyday contexts. We built a Home-like Observational Multimodal Environment (HOME) to examine how 12- to 24-month-olds and their parents jointly create input statistics for early word learning in everyday activities. Using head-mounted eye-tracking technology in the HOME lab, we will record infant gaze and parent speech during three everyday activities: toy play, book reading, and meal preparation. We will specifically focus on the moments when parents produce object names in their speech, precisely measuring the infants' in-the- moment attention toward those named objects. Through the analyses of high-density gaze and speech data, we will characterize the referential uncertainty at individual naming instances from the infant's point of view (Aim 1), examine the computational mechanisms that operate on the input statistics to achieve successful word learning (Aim 2), compare the quantity and quality of the input statistics among the three different everyday activities (Aim 3), and determine whether the quantity and quality of the input statistics in those activities predict individual differences in vocabulary growth (Aim 4). This proposed work holds significant potential for breaking barriers that stall the current understanding of early word learning, developing innovative methods to quantify the multimodal nature of the language environment, and providing mechanistic insights into the origins and consequences of individual differences in language development. Project Number: 1R01HD119490-01 | Fiscal Year: 2025 | NIH Institute/Center: Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) | Principal Investigator: Chen Yu | Institution: UNIVERSITY OF TEXAS AT AUSTIN, AUSTIN, TX | Award Amount: $344,356 | Activity Code: R01 | Study Section: Language and Communication Study Section[LCOM] View on NIH RePORTER: https://reporter.nih.gov/project-details/1R01HD11949001