Datasets for "Modal Interrogatives as Discourse-Level Constructions: Collostructional Analysis and LLM Surprisal"

This repository contains the processed datasets underlying the collostructional and surprisal analyses reported in the article Modal Interrogatives as Discourse-Level Constructions: Collostructional Analysis and LLM Surprisal. The repository includes collostructional analysis results and surprisal analysis results for subjects, verbs, and response markers in May and Might interrogatives, together with a Python script. Original corpus materials were obtained from licensed databases and are therefore not redistributed. Only the processed datasets and demonstration code necessary to reproduce the reported statistical results and figures are provided.