Large Language Models as Modal Models in Linguistics

The paper presents a philosophical framework for understanding the role of large language models (LLMs) in linguistic theory, categorizing positions into insulationism, eliminativism, and conciliationism. It argues that while LLMs serve as minimal models offering how-possibly explanations (HPEs) regarding language acquisition, they currently do not meet the criteria for how-actually explanations (HAEs) of human language. This analysis provides a nuanced perspective on the epistemic value of LLMs, aiding researchers in evaluating their application in linguistic research.