Publication

Multilingual Language Models are not Multicultural: A Case Study in Emotion

WASSA 2023 paper on cultural variation in emotional expression and multilingual language models.

Authors: Shreya Havaldar, Sunny Rai, Bhumika Singhal, Langchen Liu, Sharath Chandra Guntuku, and Lyle Ungar.

Published in the 13th Workshop on Computational Approaches to Subjectivity, Sentiment, & Social Media Analysis, 2023.

Summary

This earlier work studies whether multilingual language models capture culturally specific emotional expression across languages.

Role in my research path

This is earlier NLP and computational social-science work. It is separate from my current PDE-discovery agenda, but it reflects an early interest in evaluating what machine-learning models actually learn, rather than assuming good benchmark behavior means conceptual understanding.

Topics

  • Multilingual language models.
  • Cultural variation in emotion.
  • Social media and affective computing.

Connection to current work

The domain is different from my current scientific machine-learning work, but the evaluation theme remains relevant: models can appear competent while missing the structure that matters for the problem.

Back to homepage