Rhetorical Questions in LLM Representations: A Linear Probing Study
For NLP researchers, this work reveals that rhetorical questions are encoded via multiple linear directions in LLMs, challenging the assumption of a single shared representation.
The study uses linear probing to analyze how large language models represent rhetorical questions, finding that rhetorical signals emerge early and are linearly separable from information-seeking questions, with cross-dataset transfer achieving AUROC around 0.7-0.8. However, transferability does not imply a shared representation, as probes trained on different datasets capture distinct rhetorical phenomena.
Rhetorical questions are asked not to seek information but to persuade or signal stance. How large language models internally represent them remains unclear. We analyze rhetorical questions in LLM representations using linear probes on two social-media datasets with different discourse contexts, and find that rhetorical signals emerge early and are most stably captured by last-token representations. Rhetorical questions are linearly separable from information-seeking questions within datasets, and remain detectable under cross-dataset transfer, reaching AUROC around 0.7-0.8. However, we demonstrate that transferability does not simply imply a shared representation. Probes trained on different datasets produce different rankings when applied to the same target corpus, with overlap among the top-ranked instances often below 0.2. Qualitative analysis shows that these divergences correspond to distinct rhetorical phenomena: some probes capture discourse-level rhetorical stance embedded in extended argumentation, while others emphasize localized, syntax-driven interrogative acts. Together, these findings suggest that rhetorical questions in LLM representations are encoded by multiple linear directions emphasizing different cues, rather than a single shared direction.