Surfacing the Unsaid: CUE-Bench for Affective Stance in Chinese Discourse

2026-08-11

AI Summary

This study addresses the limitations of existing Chinese sentiment benchmarks, which predominantly focus on surface-level polarity and struggle to capture nuanced affective expressions such as implicit sentiment, irony, or pragmatic reinterpretation. To bridge this gap, we propose CUE-Bench—the first benchmark for affective stance in Chinese—defining nine interpretable stance categories through the interaction of explicit and implicit polarity. Our framework integrates pragmatic intent with fine-grained emotion annotations, establishing a multidimensional, structured approach to affective reasoning. Grounded in linguistic theory, the human-annotated dataset enables joint modeling of these dimensions. Experiments on strong baseline models demonstrate that incorporating affective stance improves fine-grained emotion recognition accuracy by 3.5% and boosts pragmatic intent detection performance by 7.8%, substantially advancing the evaluation of implicit sentiment in Chinese.

Abstract

Emotion understanding in discourse requires reasoning beyond surface sentiment because speakers often convey affect through indirect, implicit, polite, ironic, or deliberately mismatched expressions. Existing emotion benchmarks mainly annotate surface polarity or final emotion categories, while lacking a structured account of how explicit expression, implicit affect, pragmatic intent, and fine grained emotion interact. This limitation makes current evaluations insensitive to cases where affective meaning is concealed, weakened, inverted, or pragmatically reshaped, thereby obscuring model failures in deeper emotion understanding. To address this gap, we introduce CUE Bench, a Chinese Unsaid Emotion benchmark that centers on Affective Stance and covers diverse communicative scenarios. CUE Bench constructs nine human interpretable affective stances from explicit implicit polarity interaction and further provides intent and fine grained emotion annotations for structured affective inference. Experiments show that incorporating Affective Stance improves fine grained emotion recognition by 3.5 percentage points and pragmatic intent detection by 7.8 percentage points over strong baselines.

Problem

• affective stance
• implicit emotion
• pragmatic intent
• emotion understanding
• discourse

Innovation

• Affective Stance
• Implicit Emotion
• Pragmatic Intent
• Fine-grained Emotion Recognition
• CUE-Bench