LiVVaL

Linguaggio e Variazione | Variation in Language

The Consequences of Large Language Models for Linguistic Theory

crossmark logo

Abstract
Large Language Models (LLMs) have revolutionized the relationship between linguistics and artificial intelligence. This paper argues that, despite their remarkable performance, LLMs are not linguistic theories but tools that can both challenge and advance them. I will show that current models diverge from human cognition and that linguistics remains essential not only for interpreting how these systems work but for guiding the design of more cognitively plausible and interpretable language models.


open access | peer reviewed

Submitted: Nov. 10, 2025 | Accepted: Jan. 28, 2026 | Published Nov. 17, 2026 | Language: en

Keywords Artificial intelligence • Language models • Linguistic theory • Cognitive plausibility • Distributional learning


read this chapter