Project collaborators
Matthew J. Cooper Borkenhagen, Florida State University & The Florida Center for Reading Research
Christopher R. Cox, Louisiana State University
Jason Lo, Data Science Institute
Project start and end dates
10/1/2023 – 12/31/2023
Project summary
“The Data Science Institute is a premier resource for top-level scientific support for various technical work that is important in the research that I conduct. Their scientific and administrative staff are responsive and skilled, making collaboration with their team easy.”
—Matthew Cooper Borkenhagen, Assistant Professor, Florida State University
We developed a machine learning model to simulate children’s reading learning processes. This digital twin allows us to explore reading mechanisms and could lead to more personalized instruction methods in the future.
A computational model was developed using connectionist principles that include temporal processing mechanisms for orthography and phonology. The model learns to produce a phonological pattern from print for words of any length. It was trained on a large corpus of words and tested on a variety of printed words to examine model behavior relative to human readers. The model exhibits important effects associated with visual word recognition, including effects of word frequency, word consistency, and their interaction. It achieved 97% accuracy in production and can read various non-words, demonstrating its ability to generalize beyond the training set. This model captures a range of behavioral effects in reading, advancing our understanding of reading development.
Collaboration with DSI allowed me to bring elements of my research to a level that I couldn’t achieve on my own, or with my in-house research team. This included advanced knowledge of computational architectures, how to implement novel architecture into my work, and database building to support analysis. DSI is and will be an important resource for me for years to come.”
—Matthew Cooper Borkenhagen, Assistant Professor, Florida State University
Project outputs/deliverables
Presentation at the 2024 Society for the Scientific Study of Reading Annual Meeting, Copenhagen
Publication: In preparation