Toward Unsupervised Conceptual Metaphor Discovery: A Case Study in Online Immigration Discourse

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing · Depth: Expert, medium

Summary

Alexandria Leto and Maria Leonor Pacheco introduce an unsupervised, LLM-based method for discovering conceptual metaphors in text corpora, detailed in the Proceedings of the Seventh Workshop on Natural Language Processing and Computational Social Science (July 2026, pages 159–175). This approach addresses a limitation in prior computational political metaphor analysis, which typically requires predefined source and target domain inventories. Their method detects metaphorical expressions with strong performance and then clusters them to approximate source domain categories, aligning with Conceptual Metaphor Theory (CMT) where metaphors systematically map concrete source domains to abstract target domains. A proof-of-concept demonstrates its utility through a case study on online immigration discourse, showing that the generated metaphor clusters effectively provide context for frame analysis. The authors propose that these conceptual mappings can significantly guide computational political discourse analysis by revealing how different source domains frame the same target.

Key takeaway

For computational social scientists analyzing political discourse, this unsupervised LLM-based method offers a new way to uncover conceptual metaphors without predefined inventories. You can apply this technique to identify salient conceptual mappings and generate metaphor clusters, providing crucial context for frame analysis. This approach allows you to gain deeper insights into how specific topics, like immigration, are framed by different source domains.

Key insights

An unsupervised LLM-based method uncovers conceptual metaphors to enhance computational political discourse analysis.

Principles

• Metaphors systematically map concrete source to abstract target domains.
• Conceptual mappings guide computational political discourse analysis.
• Distinct source domains frame the same target in different ways.

Method

Detect metaphorical expressions from a corpus using an LLM, then cluster them to approximate source domain categories for unsupervised conceptual metaphor discovery.

In practice

• Apply LLMs for unsupervised metaphor detection in corpora.
• Utilize metaphor clusters to provide context for frame analysis.

Topics

• Conceptual Metaphor Theory
• Unsupervised Learning
• Large Language Models
• Political Discourse Analysis
• Frame Analysis
• Immigration Discourse

Best for: AI Scientist, Research Scientist, NLP Engineer

Related on AIssential

• Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing — LLMs struggle with deep metaphor comprehension, often misinterpreting concepts and relying on lexical associations over context or full syntactic understanding.
• Mapping the Political Discourse in the Brazilian Chamber of Deputies: A Multi-Faceted Computational Approach — Computational methods can reveal hidden patterns in parliamentary discourse beyond voting records.
• Graph-Augmented LLMs for Swiss MP Ideology Prediction — Graph-augmented LLMs can model complex political ideologies from text and relational data.
• From 50K to 8.2 Million in 24 Hours: Vozinha's Algorithmic Consecration and the Multilingual Making of World Cup Visibility — Algorithmic visibility for public figures is shaped by distinct cross-lingual narrative frames and platform metrics.
• A corpus-assisted metaphor analysis in portraying businesswomen: Diachronic changes in Hong Kong English news — Corpus-assisted metaphor analysis reveals diachronic shifts in media portrayals of businesswomen.
• A Resource for Enthymeme Detection in Controversial Political Discourse — The study of annotator disagreement in enthymeme detection improves model performance and understanding of human inference.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.