@inproceedings{zuo-etal-2026-speak,
title = "Speak No Evil, Just Prompt: Low-resource Multilingual Toxic Speech Detection with Audio Language Model",
author = "Zuo, Mingzi and
Zhang, Lei and
Sun, Hailiang and
Huo, Shengzhi and
Dong, Changyu and
Wang, Xin and
Wang, Bo and
Liu, Hao",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.439/",
pages = "9039--9053",
ISBN = "979-8-89176-395-1",
abstract = "The widespread dissemination of toxic content on online platforms poses a critical threat to user experience. Toxicity detection in speech receives significantly less research attention than its text counterpart. Most existing methods rely on high-resource languages and employ a cascaded pipeline combining automatic speech recognition (ASR) and text classifiers. These designs limit robustness in low-resource languages and discard important acoustic cues. To address the lack of datasets, we construct PolySpeechTox, the first toxicity-annotated speech dataset spanning 53 languages and accent varieties, with a focus on low-resource languages and multiple accents. Based on PolySpeechTox, we conduct the first systematic study of toxic speech detection under low-resource, multilingual, and multi-accent conditions. We propose SoftPrompt-TSD, a prompt-based adaptation framework that leverages a frozen audio language model to perform end-to-end toxicity detection without ASR. The decomposed soft-prompt design balances global task alignment, cross-lingual generalization, and language-specific or accent-specific calibration. On PolySpeechTox, SoftPrompt-TSD achieves a micro-averaged ROC-AUC of 98.07{\%}, mitigating the severe failures observed in baseline methods for several languages. In three generalization experiments, SoftPrompt-TSD demonstrates superior generalization capability and maintains robust performance against distribution shifts."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zuo-etal-2026-speak">
<titleInfo>
<title>Speak No Evil, Just Prompt: Low-resource Multilingual Toxic Speech Detection with Audio Language Model</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mingzi</namePart>
<namePart type="family">Zuo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hailiang</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shengzhi</namePart>
<namePart type="family">Huo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Changyu</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>The widespread dissemination of toxic content on online platforms poses a critical threat to user experience. Toxicity detection in speech receives significantly less research attention than its text counterpart. Most existing methods rely on high-resource languages and employ a cascaded pipeline combining automatic speech recognition (ASR) and text classifiers. These designs limit robustness in low-resource languages and discard important acoustic cues. To address the lack of datasets, we construct PolySpeechTox, the first toxicity-annotated speech dataset spanning 53 languages and accent varieties, with a focus on low-resource languages and multiple accents. Based on PolySpeechTox, we conduct the first systematic study of toxic speech detection under low-resource, multilingual, and multi-accent conditions. We propose SoftPrompt-TSD, a prompt-based adaptation framework that leverages a frozen audio language model to perform end-to-end toxicity detection without ASR. The decomposed soft-prompt design balances global task alignment, cross-lingual generalization, and language-specific or accent-specific calibration. On PolySpeechTox, SoftPrompt-TSD achieves a micro-averaged ROC-AUC of 98.07%, mitigating the severe failures observed in baseline methods for several languages. In three generalization experiments, SoftPrompt-TSD demonstrates superior generalization capability and maintains robust performance against distribution shifts.</abstract>
<identifier type="citekey">zuo-etal-2026-speak</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.439/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>9039</start>
<end>9053</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Speak No Evil, Just Prompt: Low-resource Multilingual Toxic Speech Detection with Audio Language Model
%A Zuo, Mingzi
%A Zhang, Lei
%A Sun, Hailiang
%A Huo, Shengzhi
%A Dong, Changyu
%A Wang, Xin
%A Wang, Bo
%A Liu, Hao
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F zuo-etal-2026-speak
%X The widespread dissemination of toxic content on online platforms poses a critical threat to user experience. Toxicity detection in speech receives significantly less research attention than its text counterpart. Most existing methods rely on high-resource languages and employ a cascaded pipeline combining automatic speech recognition (ASR) and text classifiers. These designs limit robustness in low-resource languages and discard important acoustic cues. To address the lack of datasets, we construct PolySpeechTox, the first toxicity-annotated speech dataset spanning 53 languages and accent varieties, with a focus on low-resource languages and multiple accents. Based on PolySpeechTox, we conduct the first systematic study of toxic speech detection under low-resource, multilingual, and multi-accent conditions. We propose SoftPrompt-TSD, a prompt-based adaptation framework that leverages a frozen audio language model to perform end-to-end toxicity detection without ASR. The decomposed soft-prompt design balances global task alignment, cross-lingual generalization, and language-specific or accent-specific calibration. On PolySpeechTox, SoftPrompt-TSD achieves a micro-averaged ROC-AUC of 98.07%, mitigating the severe failures observed in baseline methods for several languages. In three generalization experiments, SoftPrompt-TSD demonstrates superior generalization capability and maintains robust performance against distribution shifts.
%U https://aclanthology.org/2026.findings-acl.439/
%P 9039-9053
Markdown (Informal)
[Speak No Evil, Just Prompt: Low-resource Multilingual Toxic Speech Detection with Audio Language Model](https://aclanthology.org/2026.findings-acl.439/) (Zuo et al., Findings 2026)
ACL
- Mingzi Zuo, Lei Zhang, Hailiang Sun, Shengzhi Huo, Changyu Dong, Xin Wang, Bo Wang, and Hao Liu. 2026. Speak No Evil, Just Prompt: Low-resource Multilingual Toxic Speech Detection with Audio Language Model. In Findings of the Association for Computational Linguistics: ACL 2026, pages 9039–9053, San Diego, California, United States. Association for Computational Linguistics.