Rethinking the Idiomaticity Decomposability Hypothesis: Evidence from Distributional Learning
For linguists and NLP researchers, this work provides a computational evaluation of the decomposability hypothesis, showing it has limited predictive power for idiom behavior in distributional learning models.
The study tests the idiomaticity decomposability hypothesis using contextualized language models, finding that model-derived decomposability correlates weakly with human judgments and shows a small negative relationship with syntactic flexibility. Pretraining analyses reveal that decomposability has the strongest training-dependent effect on idiom representation stabilization, beyond frequency and surprisal.
Idioms can be analysed in terms of their decomposability, the extent to which constituent meanings contribute to the figurative whole. Decomposability is thought to predict syntactic flexibility. Usage-based accounts instead attribute idiom behaviour to distributional experience, such as speaker familiarity and predictability. We examine these views using contextualised language models as controlled distributional learners. We propose a model-internal measure of decomposability and relate it to human ratings, syntactic flexibility, and predictability while tracking idiom learning during pretraining. Model-derived decomposability correlates weakly with human judgments and shows a small but consistent negative relationship with syntactic flexibility. Pretraining analyses show that stabilisation of idiom representations in models is not explained by frequency alone. Instead, surprisal, decomposability, and frequency all contribute, with decomposability showing the strongest training-dependent effect.