Conceptual Networks for Cross-Linguistic Idiomatic Expressions:A Feature-Based Graph Approach

2026-07-13 · Source: cs.CL updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing · Depth: Expert, long

Summary

Kiran Pala and Punam Silu introduce an interpretable network-based framework for representing idiomatic and figurative meaning across eight typologically diverse languages, encompassing 160 conventional expressions. This framework annotates each expression with binary conceptual features (e.g., containment, concealment, emotional, social) derived from cognitive-linguistic theory. Pairwise Jaccard similarities define a weighted graph, revealing that idioms cluster by conceptual schema rather than by language, consistent with cognitive-linguistic predictions. The conceptual network captures unique semantic information, improves downstream idiom detection from F1 0.82 to 0.86, and remains robust when enriched with corpus frequencies. Cross-lingual transfer experiments show conceptual proximity identifies acceptable translation equivalents in 78% of cases, substantially outperforming embedding-based baselines (54%). The framework can also be scaled via automatic annotation with GPT-4, achieving 0.82 accuracy and macro-F1 0.79.

Key takeaway

For NLP Engineers developing multilingual applications, this conceptual network framework offers a robust method to handle idiomatic expressions. You should consider integrating theory-grounded conceptual features, potentially via LLM-based annotation, to improve idiom detection and facilitate accurate cross-lingual idiom mapping. This approach provides interpretability and stability beyond traditional embedding models, enhancing semantic understanding in diverse linguistic contexts.

Key insights

Idioms organize by conceptual schema, not language, offering interpretable, cross-linguistically stable meaning representations.

Principles

• Idioms cluster by conceptual schema, not by language.
• Conceptual features encode unique semantic information.
• Schemas, roles, and valence non-redundantly contribute to network structure.

Method

Manually annotate 160 idioms with binary conceptual features across three dimensions. Construct a weighted graph using Jaccard similarity. Apply community detection and evaluate against NLP tasks.

In practice

• Improve idiom detection in NLP pipelines.
• Identify cross-lingual idiom translation equivalents.
• Scale idiom annotation using LLMs like GPT-4.

Topics

• Conceptual Networks
• Idiomatic Expressions
• Cross-Linguistic Semantics
• Cognitive Linguistics
• Graph Theory
• LLM Annotation
• Idiom Detection

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• Conceptual Networks for Cross-Linguistic Idiomatic Expressions:A Feature-Based Graph Approach — Idioms cluster by conceptual schema, not language, enabling cross-lingual transfer and improved detection.
• Geometry of Semantic Space: Comparative Study of Discrete and Continuous Models — Graph-based semantic models offer more interpretable meaning organization than transformer embeddings, despite similar local topology.
• Beyond LLMs: A Linguistic Approach to Causal Graph Generation from Narrative Texts — A hybrid linguistic-LLM framework significantly improves causal graph generation from narrative texts by integrating expert linguistic features.
• A Cross-Genre Analysis of Discourse Relation Signaling in the GUM Corpus — The paper analyzes discourse relation signaling across diverse genres in the GUM Corpus.
• IdioLink: Retrieving Meaning Beyond Words Across Idiomatic and Literal Expressions — Idiom understanding in retrieval requires semantic abstraction beyond lexical overlap.
• Multilingual Idioms in Sentences and Conversations Across High-, Medium-, and Low-Resource Languages — Idiomatic expressions challenge multilingual NLP, with literal meanings and low-resource languages posing significant hurdles.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CL updates on arXiv.org.