How are logical concepts encoded in natural languages and how are they learned, processed, and understood by humans and machines?
▾
Human reasoning and communication relies on a handful of words that encode logical concepts such as and, or, no, if, some, all, the, a... . Each language has a relatively small set of such words and systematically uses them to communicate human ideas and thoughts.
My research engages with three foundational questions regarding logical words:
Learning: how are logical words learned and how do abstract logical concepts associated with these words emerge in humans and computational models such as LLMs?
Understanding and Processing: how are logical words understood and proceessed in the human mind and in computational models such as LLMs?
Crosslinguistic Representation: what are the universal and variable patterns in how logical concepts are encoded across different languages?
To address these questions I use a combination of methods including computational modeling, experimental research with human participants, and crosslinguistic formal analysis of linguistic data.