Developing a cross-cultural annotation system and metacorpus for studying infants' real world language experience

Abstract

Recent issues around reproducibility, best practices, and cultural bias impact naturalistic observational approaches as much as experimental approaches, but there has been less focus onthis area. Here, we present a new approach that leverages cross-laboratory collaborative, interdisciplinary efforts to examine important psychological questions. We illustrate this approach with a particular project that examines similarities and differences in children's early experiences with language. This project develops a comprehensive start-to-finish analysis pipeline by developing a flexible and systematic annotation system, and implementing this system across a sampling from a metacorpus of audiorecordings of diverse language communities. This resource is publicly available for use, sensitive to cultural differences, and flexible to address a variety of research questions. It is also uniquely suited for use in the development of tools for automated analysis.