@inproceedings{souganidis-etal-2026-phonologically,
title = "Phonologically-aware Automatic Speech Recognition Evaluation of Low-Resource Languages: The Case of {B}asque Dialects",
author = "Souganidis, Christoforos and
Herranz, Asier and
Saratxaga, Ibon and
Navas, Eva and
Hernaez, Inma",
editor = "Anastasopoulos, Antonis and
Markantonatou, Stella and
Ralli, Angela and
Zampieri, Marcos and
Bompolas, Stavros and
Stamou, Vivian",
booktitle = "Proceedings of the First Workshop on Dialects in {NLP} {---} A Resource Perspective",
month = may,
year = "2026",
address = "Palma de Mallorca",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.dialres-1.5/",
doi = "10.63317/262fznwr54us",
pages = "48--57",
abstract = "Automatic speech recognition models are typically trained with data of standard languages. However, their performance degrades when dealing with non-standard dialectal speech. In this paper, we present the first evaluation of an automatic speech recognition system for Basque, a low-resource language, based on spontaneous broadcast speech with high representation of dialectal speech. It relies on a 140-h manually annotated propietary corpus of television programs broadcast by Basque Radio Television, including dialect-level labels, as well as standardized and pseudo-phonetic transcriptions. We find that recognition performance significantly degrades for dialectal compared to standard speech, for all dialects present in our corpus. Subsequently, we provide a quantitative analysis of phonological phenomena based on single-word substitution errors, and identify 52 recurrent phenomena, grouped into sound deletions, epentheses, and substitutions. We further show a modest but statistically significant correlation between the number of phonological phenomena in an utterance and its recognition error rate. Our findings highlight the limitations of dialect-agnostic evaluation and motivate linguistically informed, dialect-aware strategies for automatic speech recognition in low-resource and typologically diverse languages."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="souganidis-etal-2026-phonologically">
<titleInfo>
<title>Phonologically-aware Automatic Speech Recognition Evaluation of Low-Resource Languages: The Case of Basque Dialects</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christoforos</namePart>
<namePart type="family">Souganidis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Asier</namePart>
<namePart type="family">Herranz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ibon</namePart>
<namePart type="family">Saratxaga</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eva</namePart>
<namePart type="family">Navas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Inma</namePart>
<namePart type="family">Hernaez</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 First Workshop on Dialects in NLP — A Resource Perspective</title>
</titleInfo>
<name type="personal">
<namePart type="given">Antonis</namePart>
<namePart type="family">Anastasopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stella</namePart>
<namePart type="family">Markantonatou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Angela</namePart>
<namePart type="family">Ralli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marcos</namePart>
<namePart type="family">Zampieri</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stavros</namePart>
<namePart type="family">Bompolas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivian</namePart>
<namePart type="family">Stamou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Palma de Mallorca</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Automatic speech recognition models are typically trained with data of standard languages. However, their performance degrades when dealing with non-standard dialectal speech. In this paper, we present the first evaluation of an automatic speech recognition system for Basque, a low-resource language, based on spontaneous broadcast speech with high representation of dialectal speech. It relies on a 140-h manually annotated propietary corpus of television programs broadcast by Basque Radio Television, including dialect-level labels, as well as standardized and pseudo-phonetic transcriptions. We find that recognition performance significantly degrades for dialectal compared to standard speech, for all dialects present in our corpus. Subsequently, we provide a quantitative analysis of phonological phenomena based on single-word substitution errors, and identify 52 recurrent phenomena, grouped into sound deletions, epentheses, and substitutions. We further show a modest but statistically significant correlation between the number of phonological phenomena in an utterance and its recognition error rate. Our findings highlight the limitations of dialect-agnostic evaluation and motivate linguistically informed, dialect-aware strategies for automatic speech recognition in low-resource and typologically diverse languages.</abstract>
<identifier type="citekey">souganidis-etal-2026-phonologically</identifier>
<identifier type="doi">10.63317/262fznwr54us</identifier>
<location>
<url>https://aclanthology.org/2026.dialres-1.5/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>48</start>
<end>57</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Phonologically-aware Automatic Speech Recognition Evaluation of Low-Resource Languages: The Case of Basque Dialects
%A Souganidis, Christoforos
%A Herranz, Asier
%A Saratxaga, Ibon
%A Navas, Eva
%A Hernaez, Inma
%Y Anastasopoulos, Antonis
%Y Markantonatou, Stella
%Y Ralli, Angela
%Y Zampieri, Marcos
%Y Bompolas, Stavros
%Y Stamou, Vivian
%S Proceedings of the First Workshop on Dialects in NLP — A Resource Perspective
%D 2026
%8 May
%I Association for Computational Linguistics
%C Palma de Mallorca
%F souganidis-etal-2026-phonologically
%X Automatic speech recognition models are typically trained with data of standard languages. However, their performance degrades when dealing with non-standard dialectal speech. In this paper, we present the first evaluation of an automatic speech recognition system for Basque, a low-resource language, based on spontaneous broadcast speech with high representation of dialectal speech. It relies on a 140-h manually annotated propietary corpus of television programs broadcast by Basque Radio Television, including dialect-level labels, as well as standardized and pseudo-phonetic transcriptions. We find that recognition performance significantly degrades for dialectal compared to standard speech, for all dialects present in our corpus. Subsequently, we provide a quantitative analysis of phonological phenomena based on single-word substitution errors, and identify 52 recurrent phenomena, grouped into sound deletions, epentheses, and substitutions. We further show a modest but statistically significant correlation between the number of phonological phenomena in an utterance and its recognition error rate. Our findings highlight the limitations of dialect-agnostic evaluation and motivate linguistically informed, dialect-aware strategies for automatic speech recognition in low-resource and typologically diverse languages.
%R 10.63317/262fznwr54us
%U https://aclanthology.org/2026.dialres-1.5/
%U https://doi.org/10.63317/262fznwr54us
%P 48-57
Markdown (Informal)
[Phonologically-aware Automatic Speech Recognition Evaluation of Low-Resource Languages: The Case of Basque Dialects](https://aclanthology.org/2026.dialres-1.5/) (Souganidis et al., DialRes 2026)
ACL