<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="churina-jaidka-2026-incivility">
    <titleInfo>
        <title>Incivility and Rigidity: Evaluating the Risks of Fine-Tuning LLMs for Political Argumentation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Svetlana</namePart>
        <namePart type="family">Churina</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Kokil</namePart>
        <namePart type="family">Jaidka</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>Incivility on platforms such as Twitter (now X) and Reddit complicates the development of AI systems that can support productive, rhetorically sound political argumentation. We present experiments with GPT-3.5 Turbo fine-tuned on two contrasting datasets of political discourse: high-incivility Twitter replies to U.S. Congress and low-incivility posts from Reddit’s r/ChangeMyView. Our evaluation examines how data composition and prompting strategies affect the rhetorical framing and deliberative quality of model-generated arguments. Results show that Reddit-finetuned models generate safer but rhetorically rigid arguments, while cross-platform fine-tuning amplifies adversarial tone and toxicity. Prompt-based steering reduces overt toxicity (e.g., personal attacks) but cannot fully offset the influence of noisy training data. We introduce a rhetorical evaluation rubric—covering justification, reciprocity, alignment, and authority—and provide implementation guidelines for authoring, moderation, and deliberation-support systems.</abstract>
    <identifier type="citekey">churina-jaidka-2026-incivility</identifier>
    <identifier type="doi">10.63317/5g48jjks5n35</identifier>
    <location>
        <url>https://aclanthology.org/2026.lrec-1.308/</url>
    </location>
    <part>
        <date>2026-05</date>
        <extent unit="page">
            <start>3875</start>
            <end>3883</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Incivility and Rigidity: Evaluating the Risks of Fine-Tuning LLMs for Political Argumentation
%A Churina, Svetlana
%A Jaidka, Kokil
%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 churina-jaidka-2026-incivility
%X Incivility on platforms such as Twitter (now X) and Reddit complicates the development of AI systems that can support productive, rhetorically sound political argumentation. We present experiments with GPT-3.5 Turbo fine-tuned on two contrasting datasets of political discourse: high-incivility Twitter replies to U.S. Congress and low-incivility posts from Reddit’s r/ChangeMyView. Our evaluation examines how data composition and prompting strategies affect the rhetorical framing and deliberative quality of model-generated arguments. Results show that Reddit-finetuned models generate safer but rhetorically rigid arguments, while cross-platform fine-tuning amplifies adversarial tone and toxicity. Prompt-based steering reduces overt toxicity (e.g., personal attacks) but cannot fully offset the influence of noisy training data. We introduce a rhetorical evaluation rubric—covering justification, reciprocity, alignment, and authority—and provide implementation guidelines for authoring, moderation, and deliberation-support systems.
%R 10.63317/5g48jjks5n35
%U https://aclanthology.org/2026.lrec-1.308/
%U https://doi.org/10.63317/5g48jjks5n35
%P 3875-3883

Markdown (Informal)

[Incivility and Rigidity: Evaluating the Risks of Fine-Tuning LLMs for Political Argumentation](https://aclanthology.org/2026.lrec-1.308/) (Churina & Jaidka, LREC 2026)

• Incivility and Rigidity: Evaluating the Risks of Fine-Tuning LLMs for Political Argumentation (Churina & Jaidka, LREC 2026)

ACL

• Svetlana Churina and Kokil Jaidka. 2026. Incivility and Rigidity: Evaluating the Risks of Fine-Tuning LLMs for Political Argumentation. In Proceedings of the Fifteenth Language Resources and Evaluation Conference, pages 3875–3883, Palma de Mallorca, Spain. ELRA Language Resource Association.