Language provides a precise lens into cognition and neuromotor function, and we are increasingly using machine intelligence to peer through that lens. Approximately 10% of North Americans have a communication disorder, which can originate physically (as in cerebral palsy and Parkinson’s disease), cognitively (as in aphasia and Alzheimer’s disease), or both (as in cardiovascular stroke). In this talk, I describe different technologies we’ve developed in response to these issues. I show how electromagnetic articulography can be used to augment speech recognition for people with cerebral palsy, and how by continuing deeper towards the neurological origins of speech, we can identify speech plans directly from EEG and MEG data. Deeper still, we see how syntactic and semantic differences in spontaneous speech can be indicative of cognitive disorder, and how modern tools, such as conversational robotics, can be used to extend the quality of life for people living with these conditions. I will also survey some of our current work in text informatics, including diagnosis from verbal autopsies, and online information retrieval for caregivers, in the age of post-truth.
Frank Rudzicz is a scientist at the Toronto Rehabilitation Institute (University Health Network), an assistant professor of Computer Science at the University of Toronto, co-founder and President of WinterLight Labs Inc. and President of the international joint ACL/ISCA special interest group on Speech and Language Processing for Assistive Technologies. He is the recent recipient of the Young Investigator award from the Alzheimer's Society, the Early Researcher award from the Government of Ontario, the Excellence in Applied Research award from National Speech-Language & Audiology (Canada), and a best student paper award at Interspeech 2013. His work involves machine-learning, human-computer interaction, speech-language pathology, rehabilitation engineering, signal processing, and linguistics. His work has been profiled in national and international media, including Scientific American and Wired.
Faculty Host: Alan Black