My name is Jacob Johnson, and I am a student of Computer Science and Linguistics at the University of Utah, now starting a PhD with Ana Marasović of UtahNLP.

My overarching research interest is the application of theoretical/experimental linguistics principles to natural language processing, especially through formal language theory approaches and through making neural approaches less opaque.

We are currently working on applying Instance Bundles and CondaQA, the first English reading comprehension dataset which requires reasoning about the implications of negated statements, to improve robustness of negation processing.

Last summer, I started a project in the Mathematics of Language and Cognition Lab under the direction of Aniello De Santo of the University of Utah Department of Linguistics. I am working with MITSL (subregular) grammars, implementing and evaluating grammatical inference algorithms to learn them transparently from positive data, and we are now extending this implementation to be "online" (i.e. learning from one instance at a time instead of in one batch). I also contributed to a replication of "Exposing Individuals to Foreign Accent Increases their Trust in What Nonnative Speakers Say" (Boduch-Grabka & Lev-Ari, 2021) with Rachel Hayes-Harb of the Speech Acquisition Lab, and am continuing in the Speech Lab with a study on Multi-Talker word learning!

I spent Spring 2022 semester making a relation extraction system to detect drug-drug interactions. This was a term project in CS6390 Information Extraction from Text, taught by Ellen Riloff.