Google, Inc., 309 SW 6th Ave. suite 400, Portland OR 97204
roarkbr AT SYMBOL g m a i l DOT c o m
I am a computational linguist working on various topics in natural language processing. My research interests include:
language modeling for automatic speech recognition, text entry and other applications; weighted transducers and grammars; supervised and unsupervised learning of language models; text normalization and transliteration; pronunciation modeling; text entry, accessibility and augmentative and alternative communication (AAC); syntactic parsing of text and speech; statistical models of human language processing; spoken language processing for diagnosis of neurodevelopmental and neurodegenerative disorders.
Here's a Google Research blog post about some work I was involved in, to provide transliteration keyboards in 20+ South Asian languages.
A few recent papers:
Other publications; my profile on Google scholar; Google Research page; CV.
- Tiago Pimentel, Arya D. McCarthy, Damian Blasi, Brian Roark and Ryan Cotterell. 2019. Meaning to Form: Measuring Systematicity as Information. In Proceedings of ACL, pp. 1751–1764.
- Sebastian J. Mielke, Ryan Cotterell, Kyle Gorman, Brian Roark and Jason Eisner. 2019. What Kind of Language Is Hard to Language-Model? In Proceedings of ACL, pp. 4975–4989.
- Hao Zhang, Richard Sproat, Axel H. Ng, Felix Stahlberg, Xiaochang Peng, Kyle Gorman and Brian Roark. 2019. Neural Models of Text Normalization for Speech Applications. Computational Linguistics, 45(2):293-337.