AI TrackerFrontier desk · AEST
Back to news
normalaitech

Language as Trait, Politeness as Strategy: Controlled Tests of Linguistic Effects in LLM Social-Deduction Agents

arXiv cs.CL · 7 Nov 2026, 03:31 AEST

A study on large language model agents explores how variations in language complexity and politeness affect social interactions. The findings indicate that these linguistic changes do not significantly influence trust or voting behavior among agents.

Key facts

Read aloud

A study on large language model agents explores how variations in language complexity and politeness affect social interactions. The findings indicate that these linguistic changes do not significantly influence trust or voting behavior among agents. Large language model L L M agents coordinate through conversation, and the same message can be phrased in simpler or more complex language, bluntly or politely. We test whether such differences change how other agents judge and act, using the LURK Werewolf framework, which separates each utterance's speech-act predicate from its natural-language realization. Across thirty games and four thousand three hundred thirty five replays, neither manipulation changes the trust other agents place in the speaker, and no effect on votes or night actions survives correction for multiple comparisons. In natural play, eighty eight percent of face-threatening speech is bald on-record regardless of role.

Sources in this cluster