Emily Morgan: Item-Specificity and Large Language Models

Dr. Emily Morgan is, by definition, a computational psycholinguist, so her work lies at the intersection of linguistics, psychology, and computer science. As such, she currently has a couple projects that fit these themes perfectly.

Last year, Dr. Morgan was awarded a three-year NSF grant on “Generalization versus item-specificity in language processing and change.” The project will investigate how speakers of a language use both the ability to generalize and their knowledge of specific previously-encountered items. For example, speakers know that the past tense of a novel verb glorp is glorped but the past tense of run is the irregular ran. But the relationship between these two systems remains a subject of intense debate. This project will investigate generalization and item-specificity on time scales: (1) How does online language processing recruit both generalization and item-specificity? (2) How do languages evolve to contain both generalizable structure and item-specific exceptions? In fact, this exact topic is one of the courses that Dr. Morgan taught at the 2025 LSA Summer Institute in Eugene, Oregon.

Another project that Dr. Morgan is currently overseeing is part of a larger, multicampus initiative on improvements to large language models (LLMS). The MRPI Planning Grant “Leveraging California’s Linguistic Diversity to Improve Large Language Models,” seeks to establish a cross-campus research network across six UC campuses to address the risks associated with large language models (LLMs) entrenching particular language varieties and user demographics, and the complementary potential for LLMs to accelerate and improve research diversity in the language sciences. The initiative will focus on incorporating linguistic and neurocognitive diversity into the design of language technologies. It will develop experimental protocols and pilot data, host workshops, and create online modular resources for training and curriculum development. The project will leverage the expertise of researchers in psycholinguistics, aiming to foster more equitable language technology use and expand the impact of UC’s contributions to language science. Key activities include forming research groups, conducting human subjects experiments, planning workshops, and creating shared curricular resources to enhance mentorship and research opportunities. The outcomes will include new collaborative research, increased accessibility to language technology training, and strengthened cross-campus networks.

Finally, Dr. Morgan is spearheading the linguistic department’s involvement in having UC Davis host the annual California Meeting on Psycholinguistics (CAMP[8]). More information on CAMP[8] can be found here.