@inproceedings{acquaye-etal-2026-mghana,
title = "{MG}hana-{ST}: A Low-Resource Speech Translation Dataset for Ghanaian Languages and an Analysis of Multilingual Training Trade-offs",
author = "Acquaye, Frank Lawrence Nii Adoquaye and
Parakal, Eric George and
Johnson, Jesse and
Bhimani, Kishankumar and
Basil, Jochebed Afua",
editor = "Adelani, David Ifeoluwa and
Alabi, Jesujoba and
Arnett, Catherine and
Chang, Tyler and
Downey, C.M. and
Goldman, Omer and
Li, Senyu and
Miranda, Lester James Validad and
Qi, Jirui and
Stap, David and
Winata, Genta Indra and
Zhang, Xinyu Crystina",
booktitle = "Proceedings of the 6th Workshop on Multilingual Representation Learning ({MRL} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.mrl-1.49/",
pages = "653--664",
ISBN = "979-8-89176-469-9",
abstract = "We present MGhana-ST, a speech translation dataset for four low-resource Ghanaian language varieties: Ga, Twi (Akuapem and Asante), Ewe, and Fante. MGhana-ST is an ongoing annotation effort; the experiments reported here use a fixed subset of approximately 16.1 hours of paired speech and English translation data. The audio is curated from two existing Ghanaian speech resources. The English translations in MGhana-ST are produced directly from the audio by 37 native-speaker annotators, rather than derived from source-language transcriptions as in the upstream resources, and are accompanied by verbal and non-verbal event annotations.\par Using Whisper-small as a backbone, we examine monolingual and multilingual training under severe data scarcity, reporting all baseline results as means over three training seeds. We find that flat multilingual training benefits no variety in this regime. Ga and Twi are unchanged within seed variance ($+0.51$ and $+0.06$ BLEU against monolingual standard deviations of 1.63 and 2.20), while Ewe declines by 6.99 BLEU and Fante by 5.11. The two varieties that degrade are the typologically isolated one, which is also drawn from a different source corpus, and the least-resourced one. We further compare empirical cross-lingual transfer with typology-based similarity and find that, in this four-language setting, transfer BLEU identifies which language pairs interact more closely than URIEL similarity does, although neither measure predicts which varieties benefit from joint training.\par We also report a methodological finding of independent interest. An earlier single-run version of this analysis found positive transfer for three of four varieties; that result did not survive replication across seeds. For Ga and Twi, monolingual baselines trained on two to six hours of audio have seed standard deviations of 1.63 and 2.20 BLEU, roughly five and thirty times those of the corresponding multilingual models (0.35 and 0.07). When the monolingual condition is the noisier one, a single-run comparison can show apparent transfer of this size from seed variation alone. We release MGhana-ST to support future research on African language speech technologies and low-resource multilingual speech translation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="acquaye-etal-2026-mghana">
<titleInfo>
<title>MGhana-ST: A Low-Resource Speech Translation Dataset for Ghanaian Languages and an Analysis of Multilingual Training Trade-offs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Frank</namePart>
<namePart type="given">Lawrence</namePart>
<namePart type="given">Nii</namePart>
<namePart type="given">Adoquaye</namePart>
<namePart type="family">Acquaye</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eric</namePart>
<namePart type="given">George</namePart>
<namePart type="family">Parakal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jesse</namePart>
<namePart type="family">Johnson</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kishankumar</namePart>
<namePart type="family">Bhimani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jochebed</namePart>
<namePart type="given">Afua</namePart>
<namePart type="family">Basil</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 6th Workshop on Multilingual Representation Learning (MRL 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="given">Ifeoluwa</namePart>
<namePart type="family">Adelani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jesujoba</namePart>
<namePart type="family">Alabi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Catherine</namePart>
<namePart type="family">Arnett</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tyler</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">C</namePart>
<namePart type="given">M</namePart>
<namePart type="family">Downey</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Omer</namePart>
<namePart type="family">Goldman</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Senyu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lester</namePart>
<namePart type="given">James</namePart>
<namePart type="given">Validad</namePart>
<namePart type="family">Miranda</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jirui</namePart>
<namePart type="family">Qi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Stap</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Genta</namePart>
<namePart type="given">Indra</namePart>
<namePart type="family">Winata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyu</namePart>
<namePart type="given">Crystina</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-469-9</identifier>
</relatedItem>
<abstract>We present MGhana-ST, a speech translation dataset for four low-resource Ghanaian language varieties: Ga, Twi (Akuapem and Asante), Ewe, and Fante. MGhana-ST is an ongoing annotation effort; the experiments reported here use a fixed subset of approximately 16.1 hours of paired speech and English translation data. The audio is curated from two existing Ghanaian speech resources. The English translations in MGhana-ST are produced directly from the audio by 37 native-speaker annotators, rather than derived from source-language transcriptions as in the upstream resources, and are accompanied by verbal and non-verbal event annotations.\par Using Whisper-small as a backbone, we examine monolingual and multilingual training under severe data scarcity, reporting all baseline results as means over three training seeds. We find that flat multilingual training benefits no variety in this regime. Ga and Twi are unchanged within seed variance (+0.51 and +0.06 BLEU against monolingual standard deviations of 1.63 and 2.20), while Ewe declines by 6.99 BLEU and Fante by 5.11. The two varieties that degrade are the typologically isolated one, which is also drawn from a different source corpus, and the least-resourced one. We further compare empirical cross-lingual transfer with typology-based similarity and find that, in this four-language setting, transfer BLEU identifies which language pairs interact more closely than URIEL similarity does, although neither measure predicts which varieties benefit from joint training.\par We also report a methodological finding of independent interest. An earlier single-run version of this analysis found positive transfer for three of four varieties; that result did not survive replication across seeds. For Ga and Twi, monolingual baselines trained on two to six hours of audio have seed standard deviations of 1.63 and 2.20 BLEU, roughly five and thirty times those of the corresponding multilingual models (0.35 and 0.07). When the monolingual condition is the noisier one, a single-run comparison can show apparent transfer of this size from seed variation alone. We release MGhana-ST to support future research on African language speech technologies and low-resource multilingual speech translation.</abstract>
<identifier type="citekey">acquaye-etal-2026-mghana</identifier>
<location>
<url>https://aclanthology.org/2026.mrl-1.49/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>653</start>
<end>664</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MGhana-ST: A Low-Resource Speech Translation Dataset for Ghanaian Languages and an Analysis of Multilingual Training Trade-offs
%A Acquaye, Frank Lawrence Nii Adoquaye
%A Parakal, Eric George
%A Johnson, Jesse
%A Bhimani, Kishankumar
%A Basil, Jochebed Afua
%Y Adelani, David Ifeoluwa
%Y Alabi, Jesujoba
%Y Arnett, Catherine
%Y Chang, Tyler
%Y Downey, C. M.
%Y Goldman, Omer
%Y Li, Senyu
%Y Miranda, Lester James Validad
%Y Qi, Jirui
%Y Stap, David
%Y Winata, Genta Indra
%Y Zhang, Xinyu Crystina
%S Proceedings of the 6th Workshop on Multilingual Representation Learning (MRL 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-469-9
%F acquaye-etal-2026-mghana
%X We present MGhana-ST, a speech translation dataset for four low-resource Ghanaian language varieties: Ga, Twi (Akuapem and Asante), Ewe, and Fante. MGhana-ST is an ongoing annotation effort; the experiments reported here use a fixed subset of approximately 16.1 hours of paired speech and English translation data. The audio is curated from two existing Ghanaian speech resources. The English translations in MGhana-ST are produced directly from the audio by 37 native-speaker annotators, rather than derived from source-language transcriptions as in the upstream resources, and are accompanied by verbal and non-verbal event annotations.\par Using Whisper-small as a backbone, we examine monolingual and multilingual training under severe data scarcity, reporting all baseline results as means over three training seeds. We find that flat multilingual training benefits no variety in this regime. Ga and Twi are unchanged within seed variance (+0.51 and +0.06 BLEU against monolingual standard deviations of 1.63 and 2.20), while Ewe declines by 6.99 BLEU and Fante by 5.11. The two varieties that degrade are the typologically isolated one, which is also drawn from a different source corpus, and the least-resourced one. We further compare empirical cross-lingual transfer with typology-based similarity and find that, in this four-language setting, transfer BLEU identifies which language pairs interact more closely than URIEL similarity does, although neither measure predicts which varieties benefit from joint training.\par We also report a methodological finding of independent interest. An earlier single-run version of this analysis found positive transfer for three of four varieties; that result did not survive replication across seeds. For Ga and Twi, monolingual baselines trained on two to six hours of audio have seed standard deviations of 1.63 and 2.20 BLEU, roughly five and thirty times those of the corresponding multilingual models (0.35 and 0.07). When the monolingual condition is the noisier one, a single-run comparison can show apparent transfer of this size from seed variation alone. We release MGhana-ST to support future research on African language speech technologies and low-resource multilingual speech translation.
%U https://aclanthology.org/2026.mrl-1.49/
%P 653-664
Markdown (Informal)
[MGhana-ST: A Low-Resource Speech Translation Dataset for Ghanaian Languages and an Analysis of Multilingual Training Trade-offs](https://aclanthology.org/2026.mrl-1.49/) (Acquaye et al., MRL 2026)
ACL