Innovation in Language Learning
A Prolegomena in the Relationship between Linguistics and Large Language Models: Foundations, Intersections, and Theoretical Implications
Alan Garfield, University of Dubuque (United States)
Abstract
The emergence of LLMs has raised fundamental questions about the nature of linguistic knowledge, the mechanisms underlying language acquisition, and the role of linguistic theory in computational systems. Although LLMs are trained using large scale statistical and neural methods rather than explicit grammatical rules, their development remains deeply indebted to linguistics. (Indebted but not primarily based on linguistics. In short, LLMs are engineering artifacts rooted in statistical machine learning, deep neural networks - especially the Transformer architecture - and next-token prediction trained on massive text data. Large Language Models (LLMs) such as Claude, DeepSeek-V3, Gemini, GPT-4o, GPT-4, GPT-3.5, Grok, Mistral, Phi, Qwen, and others have reshaped contemporary understanding of language, computation, and cognition. Their ability to generate coherent and contextually appropriate text has prompted renewed scrutiny of the relationship between linguistic theory and artificial intelligence LLMs challenge long standing assumptions within linguistic theory, prompting scholars to reconsider foundational concepts such as competence, performance, and the necessity of innate constraints. This essay examines that relationship, arguing that linguistics provides essential conceptual, methodological, and evaluative foundations for LLMs, while LLMs simultaneously offer new empirical tools and theoretical provocations for linguistic inquiry by exploring these intersections, drawing on recent scholarship in computational linguistics, linguistic anthropology, and cognitive science.
Keywords: AI, articial intelligence, linguistics, LLM, large language models
REFERENCES
[1] Opitz, J., Wein, S., & Schneider, N. (2025). Natural language processing RELIES on linguistics. Association for Computational Linguistics.
[2] Futrell, R., & Mahowald, K. (2026). How linguistics learned to stop worrying and love the language models. Behavioral and Brain Sciences, 49, e198.
[3] Lamoureaux, S., Castelle, M., & Weichselbraun, A. (2026). Language machines: Toward a linguistic anthropology of large language models. Journal of Linguistic Anthropology, 36, e70033.
[4] Portelance, E., & Jasbi, M. (2025). On the compatibility of generative AI and generative linguistics. Nature Computational Science.
