@article{lee-parde-2026-empathy,
    title = "Empathy Speaks in Metaphors: The Empathy-Metaphor Corpus of Figurative Language in Empathetic Text",
    author = "Lee, Gyeongeun  and
      Parde, Natalie",
    editor = "Piperidis, Stelios  and
      Bel, N{\'u}ria  and
      van den Heuvel, Henk  and
      Ide, Nancy  and
      Krek, Simon  and
      Toral, Antonio",
    journal = "International Conference on Language Resources and Evaluation",
    volume = "main",
    month = may,
    year = "2026",
    address = "Palma de Mallorca, Spain",
    publisher = "ELRA Language Resource Association",
    url = "https://preview.aclanthology.org/ingest-lrec/2026.lrec-main.873/",
    pages = "11188--11198",
    abstract = "Metaphorical language is a powerful vehicle for expressing empathy, yet it has received limited attention in computational studies of supportive communication. We introduce Empathy-Metaphor, the first corpus that explicitly annotates metaphorical spans in empathetic online peer-support. Building on 2,492 empathetic posts from an acne support forum, the dataset contains over 2,100 manually identified metaphorical spans with strong inter-annotator agreement ({\ensuremath{\kappa}}=0.85). Analyses show that metaphors are frequent, diverse, and strategically positioned, often framing acne as a battle, journey, or shared struggle. Lexical and semantic clustering highlight recurring themes of encouragement and emotional hardship, while psycholinguistic analysis emphasizes the prominence of conflict and negative emotion framings. Benchmark experiments demonstrate that transformer models, especially DeBERTa-v3, substantially outperform linear and recurrent baselines, achieving a token-level macro F1 of 0.634 and a span-level macro F1 of 0.440 under relaxed evaluation. These contributions establish a new resource for studying figurative language in empathetic text, providing insights into the creative role of metaphors in online support."
}