Language is not only about what speakers intend to say, but also about how people perceive it. A statement meant as neutral may be interpreted as rude, credible, sarcastic, or harmful depending on context and audience. While NLP has traditionally modeled language through relatively fixed operationalizations of meaning, often grounded in speaker intent, author attributes, or majority-vote annotations, these approaches can potentially overlook the diversity of reader perceptions that emerge in real-world communication.
Thus, DANIS members (Hongyu Chen and Agnieszka Faleńska) are among the co-organisers of the First Workshop on Centering Social Perception in NLP (NLPercep’26), co-located with ICWSM 2026 in Los Angeles, CA, U.S. on May 26th, to bring together researchers across NLP, computational social science, sociolinguistics, psychology, and related fields, to foster discussion about how language is perceived and how social perceptions can inform dataset design, evaluation, and the development of more ethical and inclusive language technologies.
For more information about NLPercep’26, please visit our website: https://nlpercep.github.io/workshop/
| Contact |
|---|