@inproceedings{zhang-etal-2026-culturally,
title = "Culturally-Aware Alignment for Multimodal Models via Group Reward Optimization for {S}outheast {A}sia",
author = "Zhang, Longyin and
Sun, Shuo and
He, Yingxu and
Lewis, Won Cheng Yi and
Zou, Bowei and
Aw, AiTi",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.863/",
pages = "19969--19988",
ISBN = "979-8-89176-459-0",
abstract = "Multimodal large language models (MLLMs) perform well across many tasks but often exhibit Western bias, limiting their effectiveness in regions such as Southeast Asia (SEA). In this work, we study how both evaluation and model design can be better adapted to the linguistic and cultural characteristics of this region. We present the SEA-Omni Benchmark Suite, which brings together a set of SEA-focused image and video datasets capturing phenomena such as intra-sentential code-switching and culturally grounded multimodal cues. Alongside these benchmarks, we develop MERaLiON-Omni-10B, an omni-perception model trained with a progressive, data-efficient pipeline that moves from vision-language alignment to more advanced multimodal fusion. To better account for region-specific cultural signals, we study text-mediated cultural alignment in SEA multimodal settings using GRPO to surface culturally distinctive responses that are often suppressed by generic decoding strategies. We introduce a fact-anchored dual-reward mechanism to balance cultural richness with multimodal consistency in which cultural density is measured via localized knowledge entities of generated texts, and multimodal coherence is assessed against silver textual descriptions of the multimodal input. Experimental results indicate that the resulting model performs strongly on SEA benchmarks while maintaining reasonable general capabilities."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-culturally">
<titleInfo>
<title>Culturally-Aware Alignment for Multimodal Models via Group Reward Optimization for Southeast Asia</title>
</titleInfo>
<name type="personal">
<namePart type="given">Longyin</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shuo</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yingxu</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Won</namePart>
<namePart type="given">Cheng</namePart>
<namePart type="given">Yi</namePart>
<namePart type="family">Lewis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bowei</namePart>
<namePart type="family">Zou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">AiTi</namePart>
<namePart type="family">Aw</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 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Multimodal large language models (MLLMs) perform well across many tasks but often exhibit Western bias, limiting their effectiveness in regions such as Southeast Asia (SEA). In this work, we study how both evaluation and model design can be better adapted to the linguistic and cultural characteristics of this region. We present the SEA-Omni Benchmark Suite, which brings together a set of SEA-focused image and video datasets capturing phenomena such as intra-sentential code-switching and culturally grounded multimodal cues. Alongside these benchmarks, we develop MERaLiON-Omni-10B, an omni-perception model trained with a progressive, data-efficient pipeline that moves from vision-language alignment to more advanced multimodal fusion. To better account for region-specific cultural signals, we study text-mediated cultural alignment in SEA multimodal settings using GRPO to surface culturally distinctive responses that are often suppressed by generic decoding strategies. We introduce a fact-anchored dual-reward mechanism to balance cultural richness with multimodal consistency in which cultural density is measured via localized knowledge entities of generated texts, and multimodal coherence is assessed against silver textual descriptions of the multimodal input. Experimental results indicate that the resulting model performs strongly on SEA benchmarks while maintaining reasonable general capabilities.</abstract>
<identifier type="citekey">zhang-etal-2026-culturally</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.863/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>19969</start>
<end>19988</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Culturally-Aware Alignment for Multimodal Models via Group Reward Optimization for Southeast Asia
%A Zhang, Longyin
%A Sun, Shuo
%A He, Yingxu
%A Lewis, Won Cheng Yi
%A Zou, Bowei
%A Aw, AiTi
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F zhang-etal-2026-culturally
%X Multimodal large language models (MLLMs) perform well across many tasks but often exhibit Western bias, limiting their effectiveness in regions such as Southeast Asia (SEA). In this work, we study how both evaluation and model design can be better adapted to the linguistic and cultural characteristics of this region. We present the SEA-Omni Benchmark Suite, which brings together a set of SEA-focused image and video datasets capturing phenomena such as intra-sentential code-switching and culturally grounded multimodal cues. Alongside these benchmarks, we develop MERaLiON-Omni-10B, an omni-perception model trained with a progressive, data-efficient pipeline that moves from vision-language alignment to more advanced multimodal fusion. To better account for region-specific cultural signals, we study text-mediated cultural alignment in SEA multimodal settings using GRPO to surface culturally distinctive responses that are often suppressed by generic decoding strategies. We introduce a fact-anchored dual-reward mechanism to balance cultural richness with multimodal consistency in which cultural density is measured via localized knowledge entities of generated texts, and multimodal coherence is assessed against silver textual descriptions of the multimodal input. Experimental results indicate that the resulting model performs strongly on SEA benchmarks while maintaining reasonable general capabilities.
%U https://aclanthology.org/2026.emnlp-main.863/
%P 19969-19988
Markdown (Informal)
[Culturally-Aware Alignment for Multimodal Models via Group Reward Optimization for Southeast Asia](https://aclanthology.org/2026.emnlp-main.863/) (Zhang et al., EMNLP 2026)
ACL