ChatGPT as an informant
Grammaticality judgments performed by two LLMs and 20 human subjects, on general Dutch test sentences (items 1-10), Dutch long movement in island contexts (items 21-52), and English long movement in island contexts (LLMs only, items 11-20). Human judgments collected via online survey. [analysisGPT] folder containing data file and analysis script for experiment 1 (GPT) [analysisGPTvsHumans] folder containing data file and analysis script for comparison of GPT to human performance [analysisHumans] folder containing data file and analysis script for experiment 2 (humans) (anonymous) [qualtrics_experiment_humans] folder containing the qualtrics experiment (experiment 2), including instructions and information letter containing ethical approval [dutchprompts.pdf]Dutch language prompts [testitems.pdf]Test item sets
Versions
https://doi.org/10.24416/UU01-WQBJ1N | Jun 05, 2024 |
https://doi.org/10.24416/UU01-8CJAUS | Apr 10, 2024 |
This DOI represents all versions of this publication and will resolve to the latest publication.
View contents