AI summaryⓘ
The authors examine two ideas about how language models (LMs) work in understanding language. First, they look at the idea that predicting what comes next in a sentence is key to processing language. Second, they consider the claim that big language models (LLMs) have made many psycholinguistic advances possible. They also suggest ways to combine LLMs with traditional language science models for future research.
Marr's levels of analysislanguage modelspsycholinguisticslarge language modelslanguage processingpredictioncontextlinguistic informationmodel integration
Authors
Sathvik Nair, Colin Phillips
Abstract
Under the lens of Marr's levels of analysis, we critique and extend two claims about language models (LMs) and language processing: first, that predicting upcoming linguistic information based on context is central to language processing, and second, that many advances in psycholinguistics would be impossible without large language models (LLMs). We further outline future directions that combine the strengths of LLMs with psycholinguistic models.