Amii Fellow and Canada CIFAR AI Chair Alona Fyshe has been named a co-recipient of a $2.5 million grant to study literacy in a digital and multicultural world. The project, Ensuring Full Literacy in a Multicultural and Digital World, is led by Janet Werker of the University of British Columbia with Dr. Fyshe acting as Co-Director. The multidisciplinary team of experts brings together researchers from the disciplines of psychology, computing science, linguistics, and anthropology, among others.
Over the next seven years, the project will study literacy across a variety of backdrops including language acquisition and development, bilingualism, differences in culture, and the emergence and use of new technologies such as reading on tablets or learning to read with an app.
“The majority of our current computer language models are trained from sources representing skilled language use that is comparable to a fully-fluent adult,” explains Dr. Fyshe, who is an Assistant Professor in the Faculty of Science at the University of Alberta. “But how should we train models to represent a person who is not fluent in a language, or who is acquiring their first language? My role in this grant is to create computer models to explore how language is learned in early life and in certain instances where reading ability is hindered.”
The research has implications for the study of the brain and also for possible treatments in situations where language acquisition is not proceeding as normal. Additionally, by improving understanding of how people become skilled readers, the work may also inspire new methods for training machine learning algorithms for language tasks (the project has collaborators from software companies who make language learning apps) and training machine learning models in a more general setting.
Dr. Fyshe and her team will work to create two new brain imaging datasets that will allow researchers to study representations of single-word meaning in infants and to contrast brain activity during reading between typical comprehenders and poor comprehenders. In this case, poor comprehenders are children who can read and understand single words but have trouble putting words together to understand sentences, paragraphs and beyond.
Researchers will then examine these brain imaging datasets and compare them to current computer models of language meaning to better understand the processes of language acquisition and development.
The funding, which is a Social Sciences and Humanities Research Council (SSHRC) of Canada Partnerships Grant, comes as part of the Government of Canada’s recent investment of $75 million in social sciences and humanities research, which has been awarded to more than 1,600 researchers from over 60 universities across Canada.
“I’m absolutely over the moon to be joining this stellar team of researchers” says Dr. Fyshe. “We’ve brought together a truly cross-disciplinary team who are all so passionate about language learning and literacy. It was a truly great experience just writing the grant, and now I can’t wait to get started on this important work.” A full list of awardees can be found on the SSHRC website.