<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lefever-2026-keynote">
    <titleInfo>
        <title>Keynote: The Cross-Lingual Transfer Myth: Why Modern LLMs Still Fail Without Comparable Corpora and Representations</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Els</namePart>
        <namePart type="family">Lefever</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 19th Workshop on Building and Using Comparable Corpora (BUCC)</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Reinhard</namePart>
            <namePart type="family">Rapp</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Ayla</namePart>
            <namePart type="given">Rigouts</namePart>
            <namePart type="family">Terryn</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Serge</namePart>
            <namePart type="family">Sharoff</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Pierre</namePart>
            <namePart type="family">Zweigenbaum</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>ELRA Language Resources Association (ELRA)</publisher>
            <place>
                <placeTerm type="text">Palma de Mallorca, Spain</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Comparable corpora have long served as a foundation for multilingual NLP, supporting transfer across languages in tasks such as classification, retrieval, translation, and argument mining. Yet in the era of multilingual transformers and generative models, a central question is no longer simply whether texts are comparable, but what kinds of internal representations and downstream behaviors that comparability actually enables. In this keynote, I argue that cross-lingual transfer is best understood as a continuum oscillating between shared semantic structures and language-specific realizations. Drawing on two complementary studies, I demonstrate how this tension manifests both in the data models learn from and in the representations they develop. The first case study investigates multilingual stance and argument mining using the new Russian LoveHate corpus alongside English debate data. The results indicate that translated or multilingual resources are useful but insufficient proxies for language-specific corpora: local topics, culturally situated argumentation patterns, and stance expression still shape model performance and generalization. The second case study presents a neuron-level analysis of multilingual emotion detection, showing that multilingual encoders such as XLM-R develop both polyglot neurons, which respond consistently across languages, and monolingual neurons, which remain tied to particular linguistic systems. This reveals that even successful cross-lingual emotion transfer depends on only partial internal alignment. Together, these findings suggest that multilingual NLP needs corpora that preserve culturally specific meaning while supporting robust transfer, as well as interpretability frameworks that can diagnose where multilingual systems genuinely share representations and where they merely approximate them. Comparable corpora are not just training material; they are essential to understand how cross-lingual generalization succeeds, where it breaks down, and how truly multilingual NLP can move beyond English-centric assumptions and conclusions.</abstract>
    <identifier type="citekey">lefever-2026-keynote</identifier>
    <identifier type="doi">10.63317/49d3g93xayzx</identifier>
    <location>
        <url>https://aclanthology.org/2026.bucc-1.1/</url>
    </location>
    <part>
        <date>2026-05</date>
        <detail type="page"><number>1</number></detail>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Keynote: The Cross-Lingual Transfer Myth: Why Modern LLMs Still Fail Without Comparable Corpora and Representations
%A Lefever, Els
%Y Rapp, Reinhard
%Y Terryn, Ayla Rigouts
%Y Sharoff, Serge
%Y Zweigenbaum, Pierre
%S Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC)
%D 2026
%8 May
%I ELRA Language Resources Association (ELRA)
%C Palma de Mallorca, Spain
%F lefever-2026-keynote
%X Comparable corpora have long served as a foundation for multilingual NLP, supporting transfer across languages in tasks such as classification, retrieval, translation, and argument mining. Yet in the era of multilingual transformers and generative models, a central question is no longer simply whether texts are comparable, but what kinds of internal representations and downstream behaviors that comparability actually enables. In this keynote, I argue that cross-lingual transfer is best understood as a continuum oscillating between shared semantic structures and language-specific realizations. Drawing on two complementary studies, I demonstrate how this tension manifests both in the data models learn from and in the representations they develop. The first case study investigates multilingual stance and argument mining using the new Russian LoveHate corpus alongside English debate data. The results indicate that translated or multilingual resources are useful but insufficient proxies for language-specific corpora: local topics, culturally situated argumentation patterns, and stance expression still shape model performance and generalization. The second case study presents a neuron-level analysis of multilingual emotion detection, showing that multilingual encoders such as XLM-R develop both polyglot neurons, which respond consistently across languages, and monolingual neurons, which remain tied to particular linguistic systems. This reveals that even successful cross-lingual emotion transfer depends on only partial internal alignment. Together, these findings suggest that multilingual NLP needs corpora that preserve culturally specific meaning while supporting robust transfer, as well as interpretability frameworks that can diagnose where multilingual systems genuinely share representations and where they merely approximate them. Comparable corpora are not just training material; they are essential to understand how cross-lingual generalization succeeds, where it breaks down, and how truly multilingual NLP can move beyond English-centric assumptions and conclusions.
%R 10.63317/49d3g93xayzx
%U https://aclanthology.org/2026.bucc-1.1/
%U https://doi.org/10.63317/49d3g93xayzx
%P 1

Markdown (Informal)

[Keynote: The Cross-Lingual Transfer Myth: Why Modern LLMs Still Fail Without Comparable Corpora and Representations](https://aclanthology.org/2026.bucc-1.1/) (Lefever, BUCC 2026)

• Keynote: The Cross-Lingual Transfer Myth: Why Modern LLMs Still Fail Without Comparable Corpora and Representations (Lefever, BUCC 2026)

ACL

• Els Lefever. 2026. Keynote: The Cross-Lingual Transfer Myth: Why Modern LLMs Still Fail Without Comparable Corpora and Representations. In Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC), page 1, Palma de Mallorca, Spain. ELRA Language Resources Association (ELRA).