@inproceedings{hu-etal-2026-sec,
title = "{SEC}: Selective Embedding Correction for Cross-Lingual Transfer in Low-Resource Languages",
author = "Hu, Mengting and
Xue, Wenjie and
Xu, Peiyu and
Li, Xuan and
Huang, Junpeng and
Guang, Zhiyuan and
Zhang, Xiaoru and
Gao, Hang and
Wei, Sicheng and
Fan, Bingyang and
Gao, Tiegang",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1033/",
doi = "10.18653/2026.findings-emnlp.1033",
pages = "21821--21836",
ISBN = "979-8-89176-462-0",
abstract = "Enhancing NLP for low-resource languages (LRLs) is vital for supporting linguistic diversity, yet a fundamental challenge lies in the cross-lingual semantic gap. To address this, a common approach is to initialize LRL word embeddings with those from high-resource languages, thereby enriching their semantic space. However, subsequent embedding adaptation is typically performed through uniform updates across the vocabulary, which fails to allocate larger corrections to semantically difficult or infrequent tokens (e.g., culturally specific terms in Amharic or Somali) that need them most. To address this, we propose Selective Embedding Correction (SEC), a parameter-efficient framework for dynamic semantic correction. SEC models embedding adjustments as sparse low-rank residuals regularized by a group-lasso objective. This enables data-driven selectivity that refines poorly aligned tokens while preserving well-aligned representations. Experiments on named entity recognition, part-of-speech tagging, and natural language inference across multiple low-resource languages show that SEC achieves strong improvements over baselines with only about 1{\%} additional trainable parameters."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hu-etal-2026-sec">
<titleInfo>
<title>SEC: Selective Embedding Correction for Cross-Lingual Transfer in Low-Resource Languages</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mengting</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenjie</namePart>
<namePart type="family">Xue</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peiyu</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuan</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junpeng</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyuan</namePart>
<namePart type="family">Guang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoru</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hang</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sicheng</namePart>
<namePart type="family">Wei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bingyang</namePart>
<namePart type="family">Fan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tiegang</namePart>
<namePart type="family">Gao</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>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Enhancing NLP for low-resource languages (LRLs) is vital for supporting linguistic diversity, yet a fundamental challenge lies in the cross-lingual semantic gap. To address this, a common approach is to initialize LRL word embeddings with those from high-resource languages, thereby enriching their semantic space. However, subsequent embedding adaptation is typically performed through uniform updates across the vocabulary, which fails to allocate larger corrections to semantically difficult or infrequent tokens (e.g., culturally specific terms in Amharic or Somali) that need them most. To address this, we propose Selective Embedding Correction (SEC), a parameter-efficient framework for dynamic semantic correction. SEC models embedding adjustments as sparse low-rank residuals regularized by a group-lasso objective. This enables data-driven selectivity that refines poorly aligned tokens while preserving well-aligned representations. Experiments on named entity recognition, part-of-speech tagging, and natural language inference across multiple low-resource languages show that SEC achieves strong improvements over baselines with only about 1% additional trainable parameters.</abstract>
<identifier type="citekey">hu-etal-2026-sec</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1033</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1033/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>21821</start>
<end>21836</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SEC: Selective Embedding Correction for Cross-Lingual Transfer in Low-Resource Languages
%A Hu, Mengting
%A Xue, Wenjie
%A Xu, Peiyu
%A Li, Xuan
%A Huang, Junpeng
%A Guang, Zhiyuan
%A Zhang, Xiaoru
%A Gao, Hang
%A Wei, Sicheng
%A Fan, Bingyang
%A Gao, Tiegang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F hu-etal-2026-sec
%X Enhancing NLP for low-resource languages (LRLs) is vital for supporting linguistic diversity, yet a fundamental challenge lies in the cross-lingual semantic gap. To address this, a common approach is to initialize LRL word embeddings with those from high-resource languages, thereby enriching their semantic space. However, subsequent embedding adaptation is typically performed through uniform updates across the vocabulary, which fails to allocate larger corrections to semantically difficult or infrequent tokens (e.g., culturally specific terms in Amharic or Somali) that need them most. To address this, we propose Selective Embedding Correction (SEC), a parameter-efficient framework for dynamic semantic correction. SEC models embedding adjustments as sparse low-rank residuals regularized by a group-lasso objective. This enables data-driven selectivity that refines poorly aligned tokens while preserving well-aligned representations. Experiments on named entity recognition, part-of-speech tagging, and natural language inference across multiple low-resource languages show that SEC achieves strong improvements over baselines with only about 1% additional trainable parameters.
%R 10.18653/2026.findings-emnlp.1033
%U https://aclanthology.org/2026.findings-emnlp.1033/
%U https://doi.org/10.18653/2026.findings-emnlp.1033
%P 21821-21836
Markdown (Informal)
[SEC: Selective Embedding Correction for Cross-Lingual Transfer in Low-Resource Languages](https://aclanthology.org/2026.findings-emnlp.1033/) (Hu et al., Findings 2026)
ACL
- Mengting Hu, Wenjie Xue, Peiyu Xu, Xuan Li, Junpeng Huang, Zhiyuan Guang, Xiaoru Zhang, Hang Gao, Sicheng Wei, Bingyang Fan, and Tiegang Gao. 2026. SEC: Selective Embedding Correction for Cross-Lingual Transfer in Low-Resource Languages. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 21821–21836, Budapest, Hungary. Association for Computational Linguistics.