The authors present an interpretable network-based framework for representing idiomatic and figurative meaning across eight typologically diverse languages, totaling 160 conventional expressions.
- Each expression is annotated with binary conceptual features (containment, concealment, emotional, social, etc.) derived from cognitive-linguistic theory.
- Pairwise Jaccard similarities define a weighted graph where community detection reveals clustering by conceptual schema rather than language.
- The framework captures unique semantic information not present in distributional embeddings and can be scaled via automatic annotation with LLMs.
- Cross-lingual transfer experiments show that conceptual proximity alone identifies acceptable translation equivalents across five language families, outperforming embedding-based baselines.
- Ablation studies confirm that schemas, roles, and valence contribute non-redundantly to the network's organizational properties and idiom detection performance.
The framework offers an interpretable, cross-linguistically stable representation of idiomatic meaning, combining theoretical grounding with practical utility.