To join via Zoom, use the link in the email announcement. If you haven’t received it, contact Alexandra Bogoyavlenskaya.
Abstract
Children’s language development is shaped by the quantity and quality of the speech they hear and produce in daily life. Everyday communication is highly variable and unpredictable, particularly for multilingual children, and remains difficult to capture in task-based or experimental settings. Capturing this dynamic communication requires methods that reflect the diversity of real world language use, and wearable recording technology now makes this possible.
In a longitudinal study, we collected one week of naturalistic audio recordings (~12 hours per child) from N = 331 children aged 2.9–3.3 years. We investigated how everyday communication varies across children with different language statuses (monolingual, bidialectal, bilingual, multilingual).
We built a natural language processing (NLP) pipeline that segmented transcripts into speaker utterances and words, yielding core quantitative measures (e.g., lexical diversity, mean length of utterance) for children and their interlocutors, alongside qualitative measures such as conversational balance and question behaviour. To capture multilingual dynamics, we developed a language identification system capable of detecting code-switching. By combining week-long naturalistic recordings with automated NLP methods, this approach enables scalable extraction of children’s everyday language, capturing its quantity as well as quality and multilingual dynamics.
Bio
Agnese D’Angelo is a doctoral candidate in the Department of Psychology at the University of Zurich. She is working in a large project funded by the Swiss National Science Foundation which investigates the environment of multilingual children. Her PhD is supervised by Dr. Stephanie Wermelinger, Prof Moritz Daum, and Prof Julien Mayor.