Representational Similarity Analysis

Chinese Sensorimotor and Embodiment Norms for 3,000 Lexicalized Concepts

Researchers present a large-scale normative database of sensorimotor and embodiment ratings for 3,000 Mandarin Chinese concepts, collected from 378 native speakers across 11 sensorimotor dimensions. A validation study identifies PSE-Sensorimotor and Minkowski-3 as the strongest composite predictors of lexical decision performance. An exploratory analysis finds that sensorimotor ratings are substantially recoverable from purely linguistic (distributional) representations via simple regression (mean Spearman r = .62), with visual and auditory dimensions recovering better than chemosensory ones. The work provides both a cognitive science resource and empirical evidence bearing on whether LLMs can acquire embodied conceptual knowledge from text alone.