Title: On the Influence of Discourse Connectives on the Predictions of Humans and Language Models: The Role of Event Knowledge
Auhtors: James Richard Britton, Yan Cong, Emmanuele Chersoni (Corresponding author: [email protected]), Yu-Yin Hsu and Philippe Blache
DATASETS FOLDER:
Italian and Chinese datasets used in Human and Machine behavioral analyses, with labels unified across files:
- Items.xlsx: Stimuli items, with regions and target words.
- _allReg_data.xlsx: RTs, conditions, and frequencies information.
- _LM_data.csv: Language models computed surprisal scoress.
SCRIPTS FOLDER
Python scripts
- For computing surprisals with a range of Language Models.
- Language Models are pre-trained in Italian or Chinese.
R scripts
- For constructing Linear Mixed Effects Models.
- For visualizing the pairwise comparison.