Cory Shain

I'm a linguistics PhD student at Ohio State with a computational linguistics emphasis. I work on computational models of language acquisition and human sentence processing, and I also work on the phonology of the Iyasa language of Cameroon.

Localizing incremental linguistic prediction in the mind (Skype). MIT, Department of Brain and Cognitive Sciences. May 7, 2019. [slides]
Discovering psycholinguistic effect timecourses with deconvolutional time series regression. Johns Hopkins University, Department of Cognitive Science. November 7, 2018. [slides]
Unsupervised machine learning as acquisition modeling. MIT, Department of Brain and Cognitive Sciences. August 16, 2017. [slides] [slides with visualizations (14MB)]
A sketch of the phonology and morphlogy of the Iyasa language of Cameroon. (View)
Iyasa webonary. An online interactive dictionary with audio recordings (English and French interfaces). (Vew)
2016-present -- Ph.D. in linguistics, The Ohio State University
2014-2016 -- Academic Advisor, University Exploration, The Ohio State University
2009-2014 -- Linguist/Translator, SIL International
2008-2009 -- M.A. in linguistics, The Ohio State University
2005-2008 -- B.A. in linguistics, The Ohio State University
DTSR -- I am the author of the DTSR (Deconvolutional Time Series Regression) codebase, a general purpose regression technique for modeling temporally diffuse effects in time series data (see EMNLP 2018 reference above).
DTSR -- I am the author of the DNNSeg codebase, which implements an unsupervised deep neural approach to speech segmentation and classification (see Shain and Elsner, NAACL 2019 reference above).
Unsupervised Neural Segmenter -- I am a developer (together with Micha Elsner) of an unsupervised neural speech segmenter system (see EMNLP 2017 reference above).
ModelBlocks -- I am a developer for ModelBlocks, an open-source software package for constructing and evaluating broad-coverage probabilistic models of cognitive processes.
UHHMM -- I am a contributer to the UHHMM project, an unsupervised cognitively-constrained raw-text parsing system (see COLING 2016 reference above).
VIZ -- I developed VIZ while working as an academic advisor for undergraduates. It's a responsive web tool for displaying shared requirements across academic programs, and comes with a graphical back-end management system called VIZ Builder. VIZ is written entirely in javascript, css, and html and can easily be deployed on any web server. Source code is available on Github.
2017 -- NSF Graduate Research Fellowship Program Honorable Mention
2016 -- University Fellowship, The Ohio State University