Computational Linguistics Strength
The arrival of ChatGPT in the fall of 2022 brought new attention to the field of linguistics. Scholars of the science of language have long been involved in the development of AI, particularly LLMs (large language models), and have stood to one side while their humanities colleagues are suddenly taking notice that machines were writing, and writing (sometimes) decently well.
Like all humanities disciplines, linguistics is a complicated and wide-ranging field, still shaped by the Chomskian revolution of the 1960s that emphasized the innate properties of language acquisition, while researchers have in the past decades also engaged with AI and machine learning. These “computational linguistics” researchers ponder the prospect of a final solution to the old Turing Test in creating artificial intelligence: can we build a machine so competent at using natural language that a human interlocutor, blind to the nature of the respondent, cannot tell whether that respondent is a human or a machine? At the same time, computational linguistics aims to resolve some of the oldest questions that have been asked about language acquisition.
The national reputation of the University of Utah’s computational linguistics strength is due to a trio of brilliant young scholars new to the College: assistant professors Aniello De Santo, Caleb Belth, and most recently Yang Wang, who joins the department this fall.
Our departmental focus on the fundamental properties of language will provide research to make possible new AI models for languages with a very small corpus of digitized texts. The English language corpus of training data for large language models such as Chat GPT is vast. De Santo, Belth, and Wang are interested in formal properties that make it possible to envision AI arising from even the world’s smallest linguistic communities. Their work holds promise not just for a computational reformulation of Chomsky’s paradigm about language learning, but also for a more universal approach to AI.
“The field of computational linguistics is broad: some scholars work on language technologies, from spell-checkers and spam filters to chatbots and voice assistants. Others are interested in adapting formal computational methods to study the fundamental properties of human language. While keeping an eye on recent technological advancements, our computational linguists explore the computational aspects of language as a cognitive system, with a special focus on understanding the mechanisms behind language processing and acquisition. How do humans understand and produce sounds, words, and sentences? How is human language processing tied to cognition, like memory and attention? When learning a language, given limited input, what strategies do humans adopt? What cognitive principles constrain these strategies? How are processing and acquisition related to the patterns observed in languages across the world? We begin from the interdisciplinary perspective of the cognitive sciences, integrating computational work with human experiments and ultimately drawing insights from computer science, mathematics, cognitive psychology, and linguistics."
Aniello de Santo, Ph.d.
Assistant Professor
Aniello De Santo works at the intersection of linguistics, psychology, and computer science, focusing on the role of structural representations and cognitive resources in sentence processing and morphophonological acquisition. He publishes in venues including the Proceedings of the Society for Computation in Linguistics, the ACL Workshop on Cognitive Modeling and Computational Linguistics, Language, Computational Brain & Behavior, and Language, Cognition, and Neuroscience. De Santo joined the department in 2020 after receiving a Ph.D. in linguistics from Stony Brook University and an M.S. in computer science from the University of Pavia.
Caleb Belth, Ph.d.
Assistant Professor
Caleb Belth studies computational linguistics, theoretical phonology, and language acquisition, with secondary interests in the history and philosophy of science. His research goal is to make concrete theoretical proposals about language acquisition and to work out their implications for linguistic theory using computational, experimental, and corpus based approaches. He has been awarded an NSF Graduate Research Fellowship, an NDSEG fellowship, and the University of Michigan’s Richard and Eleanor Towner Prize for Distinguished Academic Achievement. Belth joined the department in 2023; he received his Ph.D. in computer science from the University of Michigan and a B.S. in computer science from Purdue University.
Yang Wang, Ph.d.
Assistant Professor
Yang Wang focuses on theoretical phonology (sound systems) and morphophonology (how words are formed), particularly on how structural properties of natural language (morpho)phonology are represented, computed, and learned. Her most recent publication examines the computational nature of the naturally occurring copying operation as word-formation processes, such as in Ilokano (Austronesian, Philippines), where the plural of “pusa” cat is “pus-pusa” cats. Wang joined the Department of Linguistics in Fall 2024; she received both a Ph.D. in linguistics and a B.S. in linguistics and mathematics of computation from UCLA.