@inproceedings{ibrahim-2026-closing,
title = "Closing the Resource Gap for {T}igre: A Diaspora-Sourced {E}nglish{--}{T}igre Parallel Dataset for {SMOL}",
author = "Ibrahim, Beshir",
editor = "Haddow, Barry and
Kocmi, Tom and
Koehn, Philipp and
Monz, Christof",
booktitle = "Proceedings of the Eleventh Conference on Machine Translation",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.wmt-main.196/",
pages = "2510--2516",
ISBN = "979-8-89176-474-3",
abstract = "Tigre, an Afro-Semitic language maintaining strong linguistic continuity with Classical Ge{'}ez, holds a significant presence in Eritrea as well as across native and intergenerational refugee communities in eastern Sudan. It remains absent from major machine translation benchmarks and resources such as FLORES-200, despite an established orthography and active use in education. As a submission to the WMT26 Open Language Data shared task, we present a community-driven contribution of 11,998 English{--}Tigre pairs (word-, phrase-, and sentence-level) to the SMOL parallel corpus, written in the standardized Ge{'}ez-script form of Tigre used in Eritrean education and media. The data were produced through a structured, multi-stage post-editing and crossvalidation pipeline involving two diaspora native-speaker translator groups. We document the collection methodology and translator workflow, and validate the data{'}s usefulness by fine-tuning NLLB-200-3.3B on the contributed pairs, observing about a 4.4{\texttimes} improvement in chrF++ over an untrained baseline for English-to-Tigre translation (4.92 to 21.63), with gains in both translation directions. We release the dataset under a Creative Commons Attribution (CC-BY 4.0) license to support future NLP research on this severely underresourced language"
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ibrahim-2026-closing">
<titleInfo>
<title>Closing the Resource Gap for Tigre: A Diaspora-Sourced English–Tigre Parallel Dataset for SMOL</title>
</titleInfo>
<name type="personal">
<namePart type="given">Beshir</namePart>
<namePart type="family">Ibrahim</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 Eleventh Conference on Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Barry</namePart>
<namePart type="family">Haddow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tom</namePart>
<namePart type="family">Kocmi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Philipp</namePart>
<namePart type="family">Koehn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christof</namePart>
<namePart type="family">Monz</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">979-8-89176-474-3</identifier>
</relatedItem>
<abstract>Tigre, an Afro-Semitic language maintaining strong linguistic continuity with Classical Ge’ez, holds a significant presence in Eritrea as well as across native and intergenerational refugee communities in eastern Sudan. It remains absent from major machine translation benchmarks and resources such as FLORES-200, despite an established orthography and active use in education. As a submission to the WMT26 Open Language Data shared task, we present a community-driven contribution of 11,998 English–Tigre pairs (word-, phrase-, and sentence-level) to the SMOL parallel corpus, written in the standardized Ge’ez-script form of Tigre used in Eritrean education and media. The data were produced through a structured, multi-stage post-editing and crossvalidation pipeline involving two diaspora native-speaker translator groups. We document the collection methodology and translator workflow, and validate the data’s usefulness by fine-tuning NLLB-200-3.3B on the contributed pairs, observing about a 4.4× improvement in chrF++ over an untrained baseline for English-to-Tigre translation (4.92 to 21.63), with gains in both translation directions. We release the dataset under a Creative Commons Attribution (CC-BY 4.0) license to support future NLP research on this severely underresourced language</abstract>
<identifier type="citekey">ibrahim-2026-closing</identifier>
<location>
<url>https://aclanthology.org/2026.wmt-main.196/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>2510</start>
<end>2516</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Closing the Resource Gap for Tigre: A Diaspora-Sourced English–Tigre Parallel Dataset for SMOL
%A Ibrahim, Beshir
%Y Haddow, Barry
%Y Kocmi, Tom
%Y Koehn, Philipp
%Y Monz, Christof
%S Proceedings of the Eleventh Conference on Machine Translation
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-474-3
%F ibrahim-2026-closing
%X Tigre, an Afro-Semitic language maintaining strong linguistic continuity with Classical Ge’ez, holds a significant presence in Eritrea as well as across native and intergenerational refugee communities in eastern Sudan. It remains absent from major machine translation benchmarks and resources such as FLORES-200, despite an established orthography and active use in education. As a submission to the WMT26 Open Language Data shared task, we present a community-driven contribution of 11,998 English–Tigre pairs (word-, phrase-, and sentence-level) to the SMOL parallel corpus, written in the standardized Ge’ez-script form of Tigre used in Eritrean education and media. The data were produced through a structured, multi-stage post-editing and crossvalidation pipeline involving two diaspora native-speaker translator groups. We document the collection methodology and translator workflow, and validate the data’s usefulness by fine-tuning NLLB-200-3.3B on the contributed pairs, observing about a 4.4× improvement in chrF++ over an untrained baseline for English-to-Tigre translation (4.92 to 21.63), with gains in both translation directions. We release the dataset under a Creative Commons Attribution (CC-BY 4.0) license to support future NLP research on this severely underresourced language
%U https://aclanthology.org/2026.wmt-main.196/
%P 2510-2516
Markdown (Informal)
[Closing the Resource Gap for Tigre: A Diaspora-Sourced English–Tigre Parallel Dataset for SMOL](https://aclanthology.org/2026.wmt-main.196/) (Ibrahim, WMT 2026)
ACL