LinguaGame: A Linguistically Grounded Game-Theoretic Paradigm for Multi-Agent Dialogue Generation

TL;DR

LinguaGame introduces a linguistically grounded game-theoretic framework for multi-agent dialogue generation, enhancing communication efficiency by modeling dialogue as strategic signaling over intents, with minimal task-specific coupling.

Contribution

It presents a novel game-theoretic paradigm that models dialogue as signaling over intents, improving communication efficiency in multi-agent systems with minimal task-specific design.

Findings

Significant gains in communication efficiency in simulated courtroom and debate scenarios.

Effective inference-time decision adjustment using a training-free equilibrium approximation.

Framework relies on linguistically informed reasoning, reducing task-specific coupling.

Abstract

Large Language Models (LLMs) have enabled Multi-Agent Systems (MASs) where agents interact through natural language to solve complex tasks or simulate multi-party dialogues. Recent work on LLM-based MASs has mainly focused on architecture design, such as role assignment and workflow orchestration. In contrast, this paper targets the interaction process itself, aiming to improve agents' communication efficiency by helping them convey their intended meaning more effectively through language. To this end, we propose LinguaGame, a linguistically-grounded game-theoretic paradigm for multi-agent dialogue generation. Our approach models dialogue as a signalling game over communicative intents and strategies, solved with a training-free equilibrium approximation algorithm for inference-time decision adjustment. Unlike prior game-theoretic MASs, whose game designs are often tightly coupled with task-specific objectives, our framework relies on linguistically informed reasoning with minimal task-specific coupling. Specifically, it treats dialogue as intentional and strategic communication, requiring agents to infer what others aim to achieve (intents) and how they pursue those goals (strategies). We evaluate our framework in simulated courtroom proceedings and debates, with human expert assessments showing significant gains in communication efficiency.