Not What, But How: A Communicative Audit of LLM Response Framing
2026-06-01 • Computation and Language
Computation and Language
AI summaryⓘ
The authors created FRANZ, a system to evaluate how large language models (LLMs) answer subjective cultural questions, focusing not just on correctness but on how responses are framed. They also built a big dataset called SQUARE, containing many subjective questions from online discussions linked to different countries and topics. Using FRANZ, the authors found that different LLMs vary in traits like cultural viewpoint and use of human-like language when answering. They show that some traits often appear together, which helps understand differences in how LLMs present their answers.
large language modelssubjective questionsresponse framingcultural positioninganthropomorphismconversational maximsdatasetevaluation frameworkonline forumscommunicative audit
Authors
Siddhesh Milind Pawar, Sarah Masud, Haneul Yoo, Alice Oh, Isabelle Augenstein
Abstract
Large language models (LLMs) are being increasingly used to answer subjective, information-seeking questions, where users are sensitive to how responses are communicated, not just whether the answers are correct. Existing LLM evaluations for subjective cultural queries largely focus on factual correctness, ignoring how the response is framed. To this end, we introduce FRANZ, an automated FRAmework for respoNse characteriZation to conduct communicative audit of LLM responses along four dimensions: cultural positioning, use of generalizing language, anthropomorphic cues, and adherence to conversational maxims. To enable this evaluation, we contribute SQUARE - a corpus of 376k subjective questions sourced from 57 subreddits, and mapped to 7 countries and 19 question categories. We demonstrate FRANZ's applicability by scoring responses from three open-weight LLMs. We observe that LLMs show statistically significant differences in the frequency with which they employ each response characteristic. Unlike single-dimensional audits, FRANZ reveals that insider positioning and anthropomorphism are positively coupled, with the degree of coupling varying by country, providing a diagnostic lens for identifying framing divergences.