Word Prediction for AAC

Statistical language models and an Android app to help minimally verbal children with autism communicate faster.

Augmentative and alternative communication (AAC) devices help people who can’t rely on speech. Better word prediction means fewer taps per sentence.

• Stephanie Gonthier ’15 used statistical learning to improve AAC word prediction. Her talk won one of three oral-presentation awards out of 86 submissions at SVURS 2014.
• Robert Cowen ’17 built an Android AAC app that combines n-gram prediction with Brown’s stages of language development. He worked with speech-language pathologist Marissa Mitchel and Dr. Abby Hare-Harris at Geisinger, supported by CS Healthcare Engineering and Ciffolillo awards.