<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sakhaeirad-etal-2026-unmasking">
    <titleInfo>
        <title>Unmasking the Factual-Conceptual Gap in Persian Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Alireza</namePart>
        <namePart type="family">Sakhaeirad</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ali</namePart>
        <namePart type="family">Ma’manpoosh</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Arshia</namePart>
        <namePart type="family">Hemmat</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-03</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Rayyan</namePart>
            <namePart type="family">Merchant</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Karine</namePart>
            <namePart type="family">Megerdoomian</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Rabat, Morocco</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-371-5</identifier>
    </relatedItem>
    <abstract>While emerging Persian NLP benchmarks have expanded into pragmatics and politeness, they rarely distinguish between memorized cultural facts and the ability to reason about implicit social norms. We introduce DIVANBENCH, a diagnostic benchmark focused on superstitions and customs, arbitrary, context-dependent rules that resist simple logical deduction. Through 315 questions across three task types (factual retrieval, paired scenario verification, and situational reasoning), we evaluate seven Persian LLMs and reveal three critical failures: most models exhibit severe acquiescence bias, correctly identifying appropriate behaviors but failing to reject clear violations; continuous Persian pretraining amplifies this bias rather than improving reasoning, often degrading the model’s ability to discern contradictions; and all models show a 21% performance gap between retrieving factual knowledge and applying it in scenarios. These findings demonstrate that cultural competence requires more than scaling monolingual data, as current models learn to mimic cultural patterns without internalizing the underlying schemas.</abstract>
    <identifier type="citekey">sakhaeirad-etal-2026-unmasking</identifier>
    <location>
        <url>https://aclanthology.org/2026.silkroadnlp-1.1/</url>
    </location>
    <part>
        <date>2026-03</date>
        <extent unit="page">
            <start>1</start>
            <end>12</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Unmasking the Factual-Conceptual Gap in Persian Language Models
%A Sakhaeirad, Alireza
%A Ma’manpoosh, Ali
%A Hemmat, Arshia
%Y Merchant, Rayyan
%Y Megerdoomian, Karine
%S The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family
%D 2026
%8 March
%I Association for Computational Linguistics
%C Rabat, Morocco
%@ 979-8-89176-371-5
%F sakhaeirad-etal-2026-unmasking
%X While emerging Persian NLP benchmarks have expanded into pragmatics and politeness, they rarely distinguish between memorized cultural facts and the ability to reason about implicit social norms. We introduce DIVANBENCH, a diagnostic benchmark focused on superstitions and customs, arbitrary, context-dependent rules that resist simple logical deduction. Through 315 questions across three task types (factual retrieval, paired scenario verification, and situational reasoning), we evaluate seven Persian LLMs and reveal three critical failures: most models exhibit severe acquiescence bias, correctly identifying appropriate behaviors but failing to reject clear violations; continuous Persian pretraining amplifies this bias rather than improving reasoning, often degrading the model’s ability to discern contradictions; and all models show a 21% performance gap between retrieving factual knowledge and applying it in scenarios. These findings demonstrate that cultural competence requires more than scaling monolingual data, as current models learn to mimic cultural patterns without internalizing the underlying schemas.
%U https://aclanthology.org/2026.silkroadnlp-1.1/
%P 1-12

Markdown (Informal)

[Unmasking the Factual-Conceptual Gap in Persian Language Models](https://aclanthology.org/2026.silkroadnlp-1.1/) (Sakhaeirad et al., SilkRoadNLP 2026)

• Unmasking the Factual-Conceptual Gap in Persian Language Models (Sakhaeirad et al., SilkRoadNLP 2026)

ACL

• Alireza Sakhaeirad, Ali Ma'manpoosh, and Arshia Hemmat. 2026. Unmasking the Factual-Conceptual Gap in Persian Language Models. In The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family, pages 1–12, Rabat, Morocco. Association for Computational Linguistics.