Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Resolving Communicative Uncertainty through Computational Inference of Partner Intentions

Tengfei Zhang¹, Chujie Ye¹, Lusha Zhu², Chunming Lu¹; ¹Beijing Normal University, ²Peking University

Presenter: Tengfei Zhang

Effective social communication demands that individuals adeptly align their conceptual representations through the precise use of language. Yet, how individuals resolve uncertainty in selecting context-appropriate utterances remains a core question in cognitive science and a significant challenge for large language models (LLMs). In this study, 60 participants (30 same-gender dyads) performed a collaborative word generation task designed to capture the dynamics of open-ended, two-way communication. Our results show that human interlocutors can effectively resolve communicative uncertainty and achieve mutual understanding, even in unconstrained, ambiguous exchanges. Furthermore, drawing on established psycholinguistic theories, we developed computational models within the cohort-based, selection-by-competition framework to test two competing mechanisms. The findings suggest a functional division of labor: statistical learning (SL) facilitates the generation of candidate lexical cohorts, while pragmatic reasoning (PR) predominantly governs word selection within the cohort.

Topic Area: Language & Communication

Extended Abstract: Full Text PDF