This project explored the identification of idiomatic (multi-word) expressions in continuous text using large pre-trained language models, leveraging zero-shot and few-shot prompts without model fine-tuning. The evaluation focused on distinguishing literal versus idiomatic interpretations and assessing strategies for accurate idiom detection.

• EMNLP 2025 (https://aclanthology.org/2025.emnlp-main.1213/)