<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jafari-etal-2026-towards">
    <titleInfo>
        <title>Towards Reliable Evaluation of Emotional Text Generation in LLMs: Human vs. Automatic Metrics</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Sadegh</namePart>
        <namePart type="family">Jafari</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Els</namePart>
        <namePart type="family">Lefever</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Veronique</namePart>
        <namePart type="family">Hoste</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-05</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the Fifteenth Language Resources and Evaluation Conference</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Stelios</namePart>
            <namePart type="family">Piperidis</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Núria</namePart>
            <namePart type="family">Bel</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Henk</namePart>
            <namePart type="family">van den Heuvel</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Nancy</namePart>
            <namePart type="family">Ide</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Simon</namePart>
            <namePart type="family">Krek</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Antonio</namePart>
            <namePart type="family">Toral</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>ELRA Language Resource Association</publisher>
            <place>
                <placeTerm type="text">Palma de Mallorca, Spain</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Evaluating emotion generation in large language models (LLMs) remains a challenging problem due to the subjective nature of emotions and the lack of reliable automatic evaluation metrics. In this paper, we introduce a robust and extensible benchmark for systematically assessing automatic metrics in emotion generation tasks. The benchmark currently includes 13 automatic evaluation metrics and five state-of-the-art LLMs, and can be easily extended without requiring additional human annotations. Through a correlation analysis with human evaluations on a carefully curated annotated subset, we identify the emotion recognition score (ERS) metric, computed with gpt-5-nano in an oneshot setting, as the most reliable automatic evaluator, achieving a correlation exceeding 0.99. Interestingly, despite relying on the same underlying LLM, the emotion absolute score (EAS) metric shows a negative correlation, demonstrating that LLM strength alone does not guarantee automatic metric alignment with human judgment. We also provide lightweight, non-LLM-based alternatives, R2_m and R3_m, in the emotion analogy score (EAnS) metric family, suitable for low-resource settings where large models are not accessible. A comprehensive per-class emotion analysis further highlights the strengths and weaknesses of the evaluated models. Overall, our results offer a practical and scalable framework for benchmarking emotion generation evaluation metrics and pave the way for more reliable, fair, and interpretable emotional language evaluation.</abstract>
    <identifier type="citekey">jafari-etal-2026-towards</identifier>
    <identifier type="doi">10.63317/554t7yighn5u</identifier>
    <location>
        <url>https://aclanthology.org/2026.lrec-1.222/</url>
    </location>
    <part>
        <date>2026-05</date>
        <extent unit="page">
            <start>2836</start>
            <end>2847</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Towards Reliable Evaluation of Emotional Text Generation in LLMs: Human vs. Automatic Metrics
%A Jafari, Sadegh
%A Lefever, Els
%A Hoste, Veronique
%Y Piperidis, Stelios
%Y Bel, Núria
%Y van den Heuvel, Henk
%Y Ide, Nancy
%Y Krek, Simon
%Y Toral, Antonio
%S Proceedings of the Fifteenth Language Resources and Evaluation Conference
%D 2026
%8 May
%I ELRA Language Resource Association
%C Palma de Mallorca, Spain
%F jafari-etal-2026-towards
%X Evaluating emotion generation in large language models (LLMs) remains a challenging problem due to the subjective nature of emotions and the lack of reliable automatic evaluation metrics. In this paper, we introduce a robust and extensible benchmark for systematically assessing automatic metrics in emotion generation tasks. The benchmark currently includes 13 automatic evaluation metrics and five state-of-the-art LLMs, and can be easily extended without requiring additional human annotations. Through a correlation analysis with human evaluations on a carefully curated annotated subset, we identify the emotion recognition score (ERS) metric, computed with gpt-5-nano in an oneshot setting, as the most reliable automatic evaluator, achieving a correlation exceeding 0.99. Interestingly, despite relying on the same underlying LLM, the emotion absolute score (EAS) metric shows a negative correlation, demonstrating that LLM strength alone does not guarantee automatic metric alignment with human judgment. We also provide lightweight, non-LLM-based alternatives, R2_m and R3_m, in the emotion analogy score (EAnS) metric family, suitable for low-resource settings where large models are not accessible. A comprehensive per-class emotion analysis further highlights the strengths and weaknesses of the evaluated models. Overall, our results offer a practical and scalable framework for benchmarking emotion generation evaluation metrics and pave the way for more reliable, fair, and interpretable emotional language evaluation.
%R 10.63317/554t7yighn5u
%U https://aclanthology.org/2026.lrec-1.222/
%U https://doi.org/10.63317/554t7yighn5u
%P 2836-2847

Markdown (Informal)

[Towards Reliable Evaluation of Emotional Text Generation in LLMs: Human vs. Automatic Metrics](https://aclanthology.org/2026.lrec-1.222/) (Jafari et al., LREC 2026)

• Towards Reliable Evaluation of Emotional Text Generation in LLMs: Human vs. Automatic Metrics (Jafari et al., LREC 2026)

ACL

• Sadegh Jafari, Els Lefever, and Veronique Hoste. 2026. Towards Reliable Evaluation of Emotional Text Generation in LLMs: Human vs. Automatic Metrics. In Proceedings of the Fifteenth Language Resources and Evaluation Conference, pages 2836–2847, Palma de Mallorca, Spain. ELRA Language Resource Association.