Research

Capturing multiparty interactions across the developmental trajectory

with Dr. Marisa Casillas and Subin Kim (2nd-year PhD)

How does a child break into a conversation that is already underway? I work on detecting the moment a child enters multiparty talk in naturalistic daylong recordings, using linguistic alignment as the diagnostic: when speakers are engaged in the same interaction, their language converges, so a drop in alignment can mark the boundary where a child joins or leaves. The measurement builds on ALIGN 2.0, a toolkit for quantifying alignment across conversational speech, applied here to daylong caregiver-child audio.

Much of my work is the pipeline that makes this analysis possible: carrying ELAN-annotated recordings into the format the tools expect, preserving the messy realities of child speech such as sung and wordless vocalisations, and running alignment measures over the result. I will present this research in a talk at the ICIS Founding Generation Symposium in September 2026, on identifying multiparty conversation in naturalistic environments through both temporal and alignment-based measures.

Question-Elicited Speech Across Naturalistic Language Learning Environments

with Dr. Anastasia Stoops

Looking at how questions shape the speech children hear and produce across different naturalistic language learning environments. Presented at the UIUC Undergraduate Research Symposium.

↓ symposium presentation

Understanding Reading Development through Visual and Linguistic Factors

with Dr. Anastasia Stoops, Dr. Jessica Montag, and Dr. Volodymyr Kindratenko (NCSA)

Examining whether deep learning architectures (LSTMs, transformer-based models like BERT) can model and predict the eye movements of teens and adults from the visual and linguistic features of text. Scaled the fixation corpus from 225k to over 800k fixations, training on NCSA Delta GPUs.