<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="daul-etal-2026-linguistically">
    <titleInfo>
        <title>Linguistically Informed Tokenization Improves ASR for Underresourced Languages</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Massimo</namePart>
        <namePart type="given">Marie</namePart>
        <namePart type="family">Daul</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Alessio</namePart>
        <namePart type="family">Tosolini</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Claire</namePart>
        <namePart type="family">Bowern</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>Proceedings of the Fifth Workshop on NLP Applications to Field Linguistics</title>
        </titleInfo>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Rabat, Morocco</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Automatic speech recognition (ASR) is a crucial tool for linguists aiming to perform a variety of language documentation tasks. However, modern ASR systems rely on data-hungry transformer architectures, rendering them generally unusable for underresourced languages. We fine-tune a wav2vec 2.0 ASR model on Yanyhangu, an Indigenous Australian language, comparing the effects of phonemic and orthographic tokenization strategies on performance. In parallel, we explore ASR’s viability as a tool in a language documentation pipeline. We find that a linguistically informed phonemic tokenization system substantially improves word error rate (WER) and character error rate (CER) compared to a baseline orthographic tokenization scheme. Finally, we show that hand-correcting the output of an ASR model is much faster than hand-transcribing audio from scratch, demonstrating that ASR can provide significant assistance for underresourced language documentation.</abstract>
    <identifier type="citekey">daul-etal-2026-linguistically</identifier>
    <location>
        <url>https://aclanthology.org/2026.fieldmatters-1.4/</url>
    </location>
    <part>
        <date>2026-03</date>
        <extent unit="page">
            <start>31</start>
            <end>37</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Linguistically Informed Tokenization Improves ASR for Underresourced Languages
%A Daul, Massimo Marie
%A Tosolini, Alessio
%A Bowern, Claire
%S Proceedings of the Fifth Workshop on NLP Applications to Field Linguistics
%D 2026
%8 March
%I Association for Computational Linguistics
%C Rabat, Morocco
%F daul-etal-2026-linguistically
%X Automatic speech recognition (ASR) is a crucial tool for linguists aiming to perform a variety of language documentation tasks. However, modern ASR systems rely on data-hungry transformer architectures, rendering them generally unusable for underresourced languages. We fine-tune a wav2vec 2.0 ASR model on Yanyhangu, an Indigenous Australian language, comparing the effects of phonemic and orthographic tokenization strategies on performance. In parallel, we explore ASR’s viability as a tool in a language documentation pipeline. We find that a linguistically informed phonemic tokenization system substantially improves word error rate (WER) and character error rate (CER) compared to a baseline orthographic tokenization scheme. Finally, we show that hand-correcting the output of an ASR model is much faster than hand-transcribing audio from scratch, demonstrating that ASR can provide significant assistance for underresourced language documentation.
%U https://aclanthology.org/2026.fieldmatters-1.4/
%P 31-37

Markdown (Informal)

[Linguistically Informed Tokenization Improves ASR for Underresourced Languages](https://aclanthology.org/2026.fieldmatters-1.4/) (Daul et al., FieldMatters 2026)

• Linguistically Informed Tokenization Improves ASR for Underresourced Languages (Daul et al., FieldMatters 2026)

ACL

• Massimo Marie Daul, Alessio Tosolini, and Claire Bowern. 2026. Linguistically Informed Tokenization Improves ASR for Underresourced Languages. In Proceedings of the Fifth Workshop on NLP Applications to Field Linguistics, pages 31–37, Rabat, Morocco. Association for Computational Linguistics.