This dataset contains all the data required to replicate the analyses in Study 2 (experimental) of the research project titled 'A taxonomy of Interrogatives and Its Role in Human-Language Model Interactions'. This dataset comprises a total of 400 prompts, constructed as 200 matched pairs of What/How interrogatives and their Hobson’s Choice counterfactuals, based on the Belnap & Steel (1976) taxonomy of interrogative forms. Prompts were sourced from the PRISM… See the full description on the dataset page:
https://huggingface.co/datasets/carowagner/study2.
