Plausibility as Commonsense Reasoning: Humans Succeed, Large Language Models Do not

A new study reveals that large language models fail to integrate world knowledge with syntactic structure for ambiguity resolution in the same way humans do. Researchers tested Turkish language models on relative-clause attachment ambiguities and found that while humans reliably use plausibility to guide interpretation, LLMs show weak, unstable, or reversed responses to the same plausibility cues.

• →Large language models do not process syntactic ambiguities in human-like ways despite strong performance on other language tasks.
• →Human participants successfully used event plausibility to resolve attachment ambiguities in Turkish relative clauses.
• →Multiple Turkish and multilingual LLMs showed weak or incorrect responses to plausibility cues that humans found clear.
• →The findings suggest current LLMs lack reliable integration between world knowledge and syntactic processing.
• →Turkish relative-clause attachment emerges as a useful diagnostic test for evaluating language model capabilities beyond standard benchmarks.