Thomas Graf wins NSF CAREER
Professor Thomas Graf has received a CAREER award from National Science Foundation (NSF) for his project “Abstract Universals in (Morpho)Syntax: Computational Characterizations and Empirical Implications”. The CAREER award is the most prestigious honor from NSF for early-career faculty who have the potential to serve as academic role models in research and education.
This five-year CAREER proposal combines research at the intersection of generative syntax, typology, and computational linguistics with an education and outreach program to increase the computational skills of linguistics students.
The central idea is to use computational techniques to model phenomena at a higher
level of abstraction in order to unearth generalizations that do not surface clearly
at the usual fine-grained levels of description. This is particularly pronounced with
syntax and its interface to morphology, where generative accounts have become very
detailed and heterogeneous. A computational lens preserves core insights of generative
syntax while distilling them into more general principles that hold across domains
with very different technical machinery. The project fuses Minimalist grammars and
subregular complexity into a unified perspective to analyze diverse phenomena such
as island effects, first conjunct agreement, differential object marking, and cross-linguistic
limits on displacement operations. It gives third-factor explanations of typological
gaps via computational
simplicity requirements, derives new empirical predictions from these requirements, explores their ramifications for the learnability of syntactic dependencies, and identifies cognitive parallels between syntax and phonology/morphology. It thus supplements generative research with a more abstracted perspective that links it more tightly to neighboring fields.
The research directly feeds the central component of the education plan --- an open-source learning platform built on Jupyter notebooks. The platform uses linguistic phenomena and ideas to convey computational concepts to those with little mathematical background, similar to an interactive textbook with an infinite supply of exercises, examples, and illustrations. The platform's development involves both graduate and undergraduate students, and it will find application in traditional classroom settings as well as various curriculum design efforts.