Communicate to Play: Pragmatic Reasoning for Efficient Cross-Cultural Communication in Codenames

TL;DR

This paper introduces RSA+C3, a pragmatic reasoning method that improves cross-cultural communication in the game Codenames Duet by modeling cultural differences and inferring sociocultural context, leading to better collaboration.

Contribution

The paper develops RSA+C3, a novel approach combining contrastive learning and language models to address cross-cultural communication challenges in collaborative games.

Findings

RSA+C3 improves collaboration between culturally diverse players.

The models reflect culturally induced differences in common ground.

The method successfully infers sociocultural context to facilitate communication.

Abstract

Cultural differences in common ground may result in pragmatic failure and misunderstandings during communication. We develop our method Rational Speech Acts for Cross-Cultural Communication (RSA+C3) to resolve cross-cultural differences in common ground. To measure the success of our method, we study RSA+C3 in the collaborative referential game of Codenames Duet and show that our method successfully improves collaboration between simulated players of different cultures. Our contributions are threefold: (1) creating Codenames players using contrastive learning of an embedding space and LLM prompting that are aligned with human patterns of play, (2) studying culturally induced differences in common ground reflected in our trained models, and (3) demonstrating that our method RSA+C3 can ease cross-cultural communication in gameplay by inferring sociocultural context from interaction. Our code is publicly available at github.com/icwhite/codenames.