Surprisal Theory in Linguistics Deemed Tautological Without Rational Grounding

A recent paper argues that Surprisal Theory, which posits a link between linguistic unit processing difficulty and language model surprisal, is a tautology without rational grounding. The author contends that for any difficulty measure, a corresponding language model can be found, rendering the theory unfalsifiable. This ambiguity was previously masked by the assumption that the relevant language model is derived from the training corpus, an assumption now challenged by empirical findings. To overcome this, the paper suggests that the language model must be based on non-empirical factors like memory constraints or processing goals, rather than solely on behavioral data. AI

RANK_REASON The item is an academic paper discussing a theoretical concept in linguistics and computational language processing. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• Bibliographic Explorer
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• ScienceCast
• scite Smart Citations
• Surprisal Theory

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →