Wright et al., "Cognitive Linguistic Identity Fusion Score (CLIFS)" (2025)

Devin R. Wright, Jisun An, and Yong-Yeol Ahn, EMNLP (2025)
DOI | arXiv | Code

@inproceedings{wright2025clifs,
    author = {Devin R. Wright and Jisun An and Yong-Yeol Ahn},
    title = {Cognitive Linguistic Identity Fusion Score ({CLIFS}): A Scalable Cognition-Informed Approach to Quantifying Identity Fusion from Text},
    booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
    pages = {11632--11662},
    doi = {10.18653/v1/2025.emnlp-main.588},
    year = {2025},
}

CLIFS combines cognitive linguistics with large language models to measure Identity fusion—how people psychologically merge their sense of self with groups, ideologies, or brands. It delivers fully automated, scalable assessments while maintaining strong alignment with established verbal measures, surpassing existing automated methods and human annotation in benchmarks. As a practical application, the approach enhances violence risk assessment by over 240%. Inspired by Card et al. (PNAS, 2022), CLIFS uses a Masked language model to assess the similarity between the target identity and self.