<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="patil-chetia-2026-beyond-surface">
    <titleInfo>
        <title>Beyond Surface Form: Probing Script and Discourse-Marker Sensitivity in Marathi LLM Reasoning</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Tejas</namePart>
        <namePart type="family">Patil</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Barnali</namePart>
        <namePart type="family">Chetia</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>The Proceedings of the Workshop on Open Reasoning Across Cultures Languages (ORACLE)</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Leonardo</namePart>
            <namePart type="family">Ranaldi</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Giulia</namePart>
            <namePart type="family">Pucci</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Federico</namePart>
            <namePart type="family">Ranaldi</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Marco</namePart>
            <namePart type="family">Valentino</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Alexandra</namePart>
            <namePart type="family">Birch</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Fabio</namePart>
            <namePart type="given">Massimo</namePart>
            <namePart type="family">Zanzotto</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-483-5</identifier>
    </relatedItem>
    <abstract>Evaluation of multilingual large language models has expanded rapidly, yet aggregate benchmark scores often conceal how performance shifts once the surface form of a language changes. We examine this question through Marathi discourse relation classification. Five instruction-following models, Llama-3.1-8B, Llama-3.3-70B, Qwen3.6-27B, GPT-OSS-20B, and GPT-OSS-120B, are evaluated on a manually curated Marathi language dataset spanning six discourse relations, two scripts, eighteen domains, and three difficulty levels, producing 2500 model-level predictions. Qwen3.6-27B obtains the highest overall accuracy at 82.4%, followed by GPT-OSS120B at 80.8%. The central finding is a large script effect as mean accuracy falls from 78.1% on Devanagari script to 49.7% on Romanized Marathi, a decline of 28.5 percentage for content that is otherwise unchanged. Relationlevel results trace to marker form rather than presence; every instance carries an explicit marker, but Temporal and Contrastive relations use a single canonical connective and are handled reliably, while Inferential and Reformulative relations lean on multi-word phrasal markers and are frequently confused with Causal, a 20.4-point accuracy gap that holds across all five models. Llama-3.1-8B shows a more severe failure, defaulting to Contrastive across most gold relations. These results indicate that strong multilingual performance on Marathi is neither script-invariant nor uniform across discourse relations.</abstract>
    <identifier type="citekey">patil-chetia-2026-beyond-surface</identifier>
    <location>
        <url>https://aclanthology.org/2026.oracle-main.28/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>412</start>
            <end>419</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Beyond Surface Form: Probing Script and Discourse-Marker Sensitivity in Marathi LLM Reasoning
%A Patil, Tejas
%A Chetia, Barnali
%Y Ranaldi, Leonardo
%Y Pucci, Giulia
%Y Ranaldi, Federico
%Y Valentino, Marco
%Y Birch, Alexandra
%Y Zanzotto, Fabio Massimo
%S The Proceedings of the Workshop on Open Reasoning Across Cultures Languages (ORACLE)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-483-5
%F patil-chetia-2026-beyond-surface
%X Evaluation of multilingual large language models has expanded rapidly, yet aggregate benchmark scores often conceal how performance shifts once the surface form of a language changes. We examine this question through Marathi discourse relation classification. Five instruction-following models, Llama-3.1-8B, Llama-3.3-70B, Qwen3.6-27B, GPT-OSS-20B, and GPT-OSS-120B, are evaluated on a manually curated Marathi language dataset spanning six discourse relations, two scripts, eighteen domains, and three difficulty levels, producing 2500 model-level predictions. Qwen3.6-27B obtains the highest overall accuracy at 82.4%, followed by GPT-OSS120B at 80.8%. The central finding is a large script effect as mean accuracy falls from 78.1% on Devanagari script to 49.7% on Romanized Marathi, a decline of 28.5 percentage for content that is otherwise unchanged. Relationlevel results trace to marker form rather than presence; every instance carries an explicit marker, but Temporal and Contrastive relations use a single canonical connective and are handled reliably, while Inferential and Reformulative relations lean on multi-word phrasal markers and are frequently confused with Causal, a 20.4-point accuracy gap that holds across all five models. Llama-3.1-8B shows a more severe failure, defaulting to Contrastive across most gold relations. These results indicate that strong multilingual performance on Marathi is neither script-invariant nor uniform across discourse relations.
%U https://aclanthology.org/2026.oracle-main.28/
%P 412-419

Markdown (Informal)

[Beyond Surface Form: Probing Script and Discourse-Marker Sensitivity in Marathi LLM Reasoning](https://aclanthology.org/2026.oracle-main.28/) (Patil & Chetia, ORACLE 2026)

• Beyond Surface Form: Probing Script and Discourse-Marker Sensitivity in Marathi LLM Reasoning (Patil & Chetia, ORACLE 2026)

ACL

• Tejas Patil and Barnali Chetia. 2026. Beyond Surface Form: Probing Script and Discourse-Marker Sensitivity in Marathi LLM Reasoning. In The Proceedings of the Workshop on Open Reasoning Across Cultures Languages (ORACLE), pages 412–419, Budapest, Hungary. Association for Computational Linguistics.