Grant: $452,151 - National Science Foundation - Aug. 19, 2009
No votes have been cast for this award yet
Award Description: Toddler word learning is a remarkable phenomenon. Starting from scratch, children progress knowing one or two words to several thousands in several short years. Most theories of learning have focused on macro level descriptions and folk-psychological constructs, conceptualizing the phenomenon in terms of the mapping of words to referents, inferences about speaker’s intentions, and joint attention (e.g. Bloom, 2000). These descriptions may capture higher level regularities, but they fall far short of a mechanistic account of how word learning works in real time. Toddlers learn words through millisecond by millisecond, second by second, and minute by minute sensorimotor events that are generated by actively engaging in the world, with objects, and with their social partners who offer object names, gestures and actions. Our preliminary evidence suggests that the smoothness of these dynamic couplings -- both within the sensorimotor system of the individual and across the coupled dyad -- are critical components of toddlers’ prowess in word learning in the cluttered and noisy contexts of everyday life. The proposed research seeks to describe these dependencies and to discover how they organize word learning in toddlers. The proposed research is transformative in the following ways: Method. We will measure the dynamic multimodal dependencies within and across social partners as they actively engage with and talk about objects. We will collect multiple streams of real-time sensorimotor data from both participants --- the first-person view of the events via tiny head cameras, hand, head and body movements via motion sensors and an array of video cameras, and spoken words via audio recording. The method breaks a barrier in the collection of real-time sensorimotor data in a naturalistic learning context. The method is also tested and has been shown to generate new data on sensorimotor dependencies between the participants that predict successful learning. Data analysis and data mining: The dense and rich streams of multimodal data are useful only to the degree that we can find meaningful patterns in those dynamic streams that bring new insights into real–time learning events. To this end, we have already made significant preliminary progress in developing new methods of analysis, visualization and data mining. These are of clear importance to understanding the real-time dynamics of toddler word learning; these new methods also have broad and transformative applications in other domains. A large amount of high-resolution high-quality multimedia data (video and audio, etc.) has been collected in social and behavioral studies. How to automatically and effectively discover new knowledge from rich multimedia data poses a compelling challenge. Learning and interaction as coupled complex systems. Each moment of sensorimotor activity by the learner determines the next – a head turn determines what is seen next which may determine what is reached for and brought close to the eyes which selects and generates the next view. We will measure how information is exchanged between the two partners, between the two multimodal systems, how do the actions of each participant organize the dynamic pattern of the learning event, and how does this contribute to learning itself? To address these questions, we will quantify fine-grained behavioral patterns within an individual’s sensorimotor system and across social partners. This constitutes a significant advance in theoretical approaches to early word learning and one that also has broad application. Measuring interaction patterns within and between complex systems is a critical problem across science – from cells, to brains, to coupled physical systems, to human-computer interaction, to groups of animals, to teams of people. Thus, this research will bring new methods and analytic tools for measuring the information in coupled interactive systems.
Project Description: We will collect multimodal and multi-streaming sensorimotor data from everyday child-parent interaction, and we will use data mining techniques to analyze the data and discover novel interaction patterns that lead to successful word learning.
Jobs Summary: Not Started (Total jobs reported: 0)
Project Status: Not Started
This award's data was last updated on Aug. 19, 2009. Help expand these official descriptions using the wiki below.