The Frequency Confound in Language-Model Surprisal and Metaphor Novelty
For researchers using LM surprisal as a proxy for cognitive processing, this work highlights a confound that undermines the validity of such measures for metaphor novelty.
The study shows that lexical frequency is a stronger predictor of metaphor novelty than LM surprisal, and the surprisal-novelty association peaks early in training, mirroring the surprisal-frequency association. This suggests that reported links between surprisal and metaphor novelty may be confounded by frequency.
Language-model (LM) surprisal is widely used as a proxy for contextual predictability and has been reported to correlate with metaphor novelty judgments. However, surprisal is tightly intertwined with lexical frequency. We explore this interaction on metaphor novelty ratings using two different word frequency measures. We analyse surprisal estimates from eight Pythia model sizes and 154 training checkpoints. Across settings, word frequency is a stronger predictor of metaphor novelty than surprisal. Across training stages, the surprisal--novelty association peaks at an early stage and then falls again, mirroring a similarly timed increase in the surprisal--frequency association. These results suggest that the often-reported optimal LM surprisal settings may incorrectly associate contextual predictability with metaphor novelty and processing difficulty, whereas lexical frequency may be the major underlying factor.