<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-dredze-2020-languages">
    <titleInfo>
        <title>Are All Languages Created Equal in Multilingual BERT?</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Shijie</namePart>
        <namePart type="family">Wu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Mark</namePart>
        <namePart type="family">Dredze</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2020-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 5th Workshop on Representation Learning for NLP</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Spandana</namePart>
            <namePart type="family">Gella</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Johannes</namePart>
            <namePart type="family">Welbl</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Marek</namePart>
            <namePart type="family">Rei</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Fabio</namePart>
            <namePart type="family">Petroni</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Patrick</namePart>
            <namePart type="family">Lewis</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Emma</namePart>
            <namePart type="family">Strubell</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Minjoon</namePart>
            <namePart type="family">Seo</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Hannaneh</namePart>
            <namePart type="family">Hajishirzi</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Online</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Multilingual BERT (mBERT) trained on 104 languages has shown surprisingly good cross-lingual performance on several NLP tasks, even without explicit cross-lingual signals. However, these evaluations have focused on cross-lingual transfer with high-resource languages, covering only a third of the languages covered by mBERT. We explore how mBERT performs on a much wider set of languages, focusing on the quality of representation for low-resource languages, measured by within-language performance. We consider three tasks: Named Entity Recognition (99 languages), Part-of-speech Tagging and Dependency Parsing (54 languages each). mBERT does better than or comparable to baselines on high resource languages but does much worse for low resource languages. Furthermore, monolingual BERT models for these languages do even worse. Paired with similar languages, the performance gap between monolingual BERT and mBERT can be narrowed. We find that better models for low resource languages require more efficient pretraining techniques or more data.</abstract>
    <identifier type="citekey">wu-dredze-2020-languages</identifier>
    <identifier type="doi">10.18653/v1/2020.repl4nlp-1.16</identifier>
    <location>
        <url>https://aclanthology.org/2020.repl4nlp-1.16</url>
    </location>
    <part>
        <date>2020-07</date>
        <extent unit="page">
            <start>120</start>
            <end>130</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Are All Languages Created Equal in Multilingual BERT?
%A Wu, Shijie
%A Dredze, Mark
%Y Gella, Spandana
%Y Welbl, Johannes
%Y Rei, Marek
%Y Petroni, Fabio
%Y Lewis, Patrick
%Y Strubell, Emma
%Y Seo, Minjoon
%Y Hajishirzi, Hannaneh
%S Proceedings of the 5th Workshop on Representation Learning for NLP
%D 2020
%8 July
%I Association for Computational Linguistics
%C Online
%F wu-dredze-2020-languages
%X Multilingual BERT (mBERT) trained on 104 languages has shown surprisingly good cross-lingual performance on several NLP tasks, even without explicit cross-lingual signals. However, these evaluations have focused on cross-lingual transfer with high-resource languages, covering only a third of the languages covered by mBERT. We explore how mBERT performs on a much wider set of languages, focusing on the quality of representation for low-resource languages, measured by within-language performance. We consider three tasks: Named Entity Recognition (99 languages), Part-of-speech Tagging and Dependency Parsing (54 languages each). mBERT does better than or comparable to baselines on high resource languages but does much worse for low resource languages. Furthermore, monolingual BERT models for these languages do even worse. Paired with similar languages, the performance gap between monolingual BERT and mBERT can be narrowed. We find that better models for low resource languages require more efficient pretraining techniques or more data.
%R 10.18653/v1/2020.repl4nlp-1.16
%U https://aclanthology.org/2020.repl4nlp-1.16
%U https://doi.org/10.18653/v1/2020.repl4nlp-1.16
%P 120-130

Markdown (Informal)

[Are All Languages Created Equal in Multilingual BERT?](https://aclanthology.org/2020.repl4nlp-1.16) (Wu & Dredze, RepL4NLP 2020)

• Are All Languages Created Equal in Multilingual BERT? (Wu & Dredze, RepL4NLP 2020)

ACL

• Shijie Wu and Mark Dredze. 2020. Are All Languages Created Equal in Multilingual BERT?. In Proceedings of the 5th Workshop on Representation Learning for NLP, pages 120–130, Online. Association for Computational Linguistics.