<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-deng-2026-nrd">
    <titleInfo>
        <title>NRD: A Hybrid Disentanglement Framework for Mitigating Interference in Multilingual Machine Translation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Jiarui</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yifan</namePart>
        <namePart type="family">Deng</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>Negative interference from cross-lingual conflicting syntactic patterns is a primary obstacle in Multilingual Neural Machine Translation (MNMT). We trace this problem to the entanglement of transferable, universal semantics with non-transferable, language-specific syntactic structures. Existing methods, relying on disjoint training-only specialization or inference-only filtering, fail to fully resolve this fundamental entanglement. To address this, we propose NRD (Neuron Representation Disentanglement), a two-stage hybrid framework that couples training-time specialization with inference-time filtering. First, a Specialization Fine-tuning stage identifies functional neurons via a semantic-invariant activation-variance metric and reinforces intrinsic modularity through sparse updates. Second, a Dynamic Representation Filtering stage purifies semantic representations at inference by adaptively suppressing syntax-sensitive neurons, guided by each language’s pre-computed gradient consistency. On the OPUS-100 benchmark, NRD outperforms strong baselines, achieving an average gain of +1.9 BLEU on supervised directions. On the WMT-10 zero-shot benchmark, it obtains a substantial +7.1 BLEU, demonstrating robust cross-lingual generalization. These results provide strong evidence that our hybrid approach effectively purifies semantic representations by mitigating syntactic interference, paving the way for more robust cross-lingual generalization.</abstract>
    <identifier type="citekey">zhang-deng-2026-nrd</identifier>
    <identifier type="doi">10.63317/55wnhwvmezwx</identifier>
    <location>
        <url>https://aclanthology.org/2026.lrec-1.677/</url>
    </location>
    <part>
        <date>2026-05</date>
        <extent unit="page">
            <start>8577</start>
            <end>8586</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T NRD: A Hybrid Disentanglement Framework for Mitigating Interference in Multilingual Machine Translation
%A Zhang, Jiarui
%A Deng, Yifan
%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 zhang-deng-2026-nrd
%X Negative interference from cross-lingual conflicting syntactic patterns is a primary obstacle in Multilingual Neural Machine Translation (MNMT). We trace this problem to the entanglement of transferable, universal semantics with non-transferable, language-specific syntactic structures. Existing methods, relying on disjoint training-only specialization or inference-only filtering, fail to fully resolve this fundamental entanglement. To address this, we propose NRD (Neuron Representation Disentanglement), a two-stage hybrid framework that couples training-time specialization with inference-time filtering. First, a Specialization Fine-tuning stage identifies functional neurons via a semantic-invariant activation-variance metric and reinforces intrinsic modularity through sparse updates. Second, a Dynamic Representation Filtering stage purifies semantic representations at inference by adaptively suppressing syntax-sensitive neurons, guided by each language’s pre-computed gradient consistency. On the OPUS-100 benchmark, NRD outperforms strong baselines, achieving an average gain of +1.9 BLEU on supervised directions. On the WMT-10 zero-shot benchmark, it obtains a substantial +7.1 BLEU, demonstrating robust cross-lingual generalization. These results provide strong evidence that our hybrid approach effectively purifies semantic representations by mitigating syntactic interference, paving the way for more robust cross-lingual generalization.
%R 10.63317/55wnhwvmezwx
%U https://aclanthology.org/2026.lrec-1.677/
%U https://doi.org/10.63317/55wnhwvmezwx
%P 8577-8586

Markdown (Informal)

[NRD: A Hybrid Disentanglement Framework for Mitigating Interference in Multilingual Machine Translation](https://aclanthology.org/2026.lrec-1.677/) (Zhang & Deng, LREC 2026)

• NRD: A Hybrid Disentanglement Framework for Mitigating Interference in Multilingual Machine Translation (Zhang & Deng, LREC 2026)

ACL

• Jiarui Zhang and Yifan Deng. 2026. NRD: A Hybrid Disentanglement Framework for Mitigating Interference in Multilingual Machine Translation. In Proceedings of the Fifteenth Language Resources and Evaluation Conference, pages 8577–8586, Palma de Mallorca, Spain. ELRA Language Resource Association.