Document Type
Article
Major
Psychological Sciences
Mentor
Prof. Harry van der Hulst, Dept. of Linguistics
Disciplines
First and Second Language Acquisition | Linguistics | Other Linguistics | Social and Behavioral Sciences
Abstract
Artificial Intelligence (AI) and Large Language Models (LLMs) have revolutionized linguistic theory, raising questions about human language acquisition. Some argue AI's language generation challenges the concept of innate linguistic ability, while others contend it reinforces it. This research examines AI's "Surplus of the Stimulus" learning—relying on vast datasets—against human language acquisition, particularly within Chomskyan frameworks of Universal Grammar and the "Poverty of the Stimulus." By analyzing language production across three groups—Native English Speakers, Post-Critical Period Second Language Learners, and LLMs—this study compares linguistic complexity using the Lexical Opulence Index (LOI). Data is collected through structured prompts across four linguistic registers: literary, simplified, technical, and speculative. Findings will clarify the mechanisms of human language learning, address AI’s limitations in true linguistic acquisition, and inform second-language learning methodologies. This research underscores the fundamental distinctions between AI-generated and human language, supporting theories of innate linguistic structures.
Recommended Citation
van Rijnswou, Hannah, "Lost in Translation: Chomskyan Analysis of Language Variation in AI, Post-Critical Learners, and Native Speakers" (2025). Holster Scholar Projects. 67.
https://digitalcommons.lib.uconn.edu/srhonors_holster/67
Comments
I would like to express my overwhelming gratitude for Mr. and Mrs. Holster, whose funding makes it possible for myself and others to pursue their undergraduate research interests, and to begin finding their place in the world of research. I would also like to thank Dr. Moscardelli and the Holster Scholars Program, who supported me continuously throughout this process as my project finally took shape from an idea to a finished product.
I also want to sincerely thank Dr. van der Hulst, who planted the seed for my love of linguistics, and, in acting as my mentor throughout this process, allowed this idea to flourish. I could not have accomplished this project without his guidance and support.
Finally, I want to thank the other 2025 Holster Scholars for building a community where we could reach out to one another for anything during the research process.