Dual Alignment Between Language Model Layers and Human Sentence Processing

ArXi:2604.18563v1 Announce Type: new A recent study (Kuribayashi, 2025) has shown that human sentence processing behavior, typically measured on syntactically unchallenging constructions, can be effectively modeled using surprisal from early layers of large language models (LLMs). This raises the question of whether such advantages of internal layers extend to syntactically challenging constructions, where surprisal has been reported to underestimate human cognitive effort.