@inproceedings{chennuru-adebayo-2026-neutral,
title = "When Neutral Turns Negative: Cross-Domain Failure Modes in {H}inglish Political Sentiment Analysis",
author = "Chennuru, Rahul and
Adebayo, Kolawole John",
editor = "Afli, Haithem and
Bouamor, Houda and
Zaghouani, Wajdi and
Ghannay, Sahar and
Hossain, Shehenaz",
booktitle = "Proceedings of the 3rd Workshop on Natural Language Processing for Political Sciences ({P}olitical{NLP} 2026)",
month = may,
year = "2026",
address = "Palma, Mallorca (Spain)",
publisher = "ELRA Language Resources Association (ELRA)",
url = "https://aclanthology.org/2026.politicalnlp-1.24/",
doi = "10.63317/3evceg5aip7m",
pages = "219--227",
abstract = "Sentiment analysis models are increasingly deployed to analyze political discourse, yet strong in-domain performance does not guarantee robustness under domain shift. We study cross-domain generalization in Hinglish (Hindi{--}English code-mixed) sentiment analysis by evaluating a fine-tuned XLM-RoBERTa classifier, trained on 29,000 general-domain Hinglish sentences, on a curated benchmark of politically oriented Hinglish text. While the model achieves 92.02{\%} accuracy in-domain, performance drops to 71.83{\%} under political domain shift. Error analysis reveals a pronounced directional bias with 48.9{\%} of neutral political statements misclassified as negative, indicating a systematic neutrality-to-negative shift. In addition, 87.5{\%} of incorrect predictions are assigned confidence scores above 95{\%}, pointing to severe miscalibration under distribution shift. We further compare these results against an instruction-tuned large language model (Llama 3.3), which achieves 90.85{\%} zero-shot accuracy and 94.37{\%} accuracy with contextual prompting, while substantially reducing neutrality bias. Our findings indicate the need for domain-aware evaluation, calibration diagnostics, and explicit reporting of failure modes when deploying sentiment models in politically sensitive settings."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chennuru-adebayo-2026-neutral">
<titleInfo>
<title>When Neutral Turns Negative: Cross-Domain Failure Modes in Hinglish Political Sentiment Analysis</title>
</titleInfo>
<name type="personal">
<namePart type="given">Rahul</namePart>
<namePart type="family">Chennuru</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kolawole</namePart>
<namePart type="given">John</namePart>
<namePart type="family">Adebayo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 3rd Workshop on Natural Language Processing for Political Sciences (PoliticalNLP 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Haithem</namePart>
<namePart type="family">Afli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Houda</namePart>
<namePart type="family">Bouamor</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wajdi</namePart>
<namePart type="family">Zaghouani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sahar</namePart>
<namePart type="family">Ghannay</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shehenaz</namePart>
<namePart type="family">Hossain</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA Language Resources Association (ELRA)</publisher>
<place>
<placeTerm type="text">Palma, Mallorca (Spain)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Sentiment analysis models are increasingly deployed to analyze political discourse, yet strong in-domain performance does not guarantee robustness under domain shift. We study cross-domain generalization in Hinglish (Hindi–English code-mixed) sentiment analysis by evaluating a fine-tuned XLM-RoBERTa classifier, trained on 29,000 general-domain Hinglish sentences, on a curated benchmark of politically oriented Hinglish text. While the model achieves 92.02% accuracy in-domain, performance drops to 71.83% under political domain shift. Error analysis reveals a pronounced directional bias with 48.9% of neutral political statements misclassified as negative, indicating a systematic neutrality-to-negative shift. In addition, 87.5% of incorrect predictions are assigned confidence scores above 95%, pointing to severe miscalibration under distribution shift. We further compare these results against an instruction-tuned large language model (Llama 3.3), which achieves 90.85% zero-shot accuracy and 94.37% accuracy with contextual prompting, while substantially reducing neutrality bias. Our findings indicate the need for domain-aware evaluation, calibration diagnostics, and explicit reporting of failure modes when deploying sentiment models in politically sensitive settings.</abstract>
<identifier type="citekey">chennuru-adebayo-2026-neutral</identifier>
<identifier type="doi">10.63317/3evceg5aip7m</identifier>
<location>
<url>https://aclanthology.org/2026.politicalnlp-1.24/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>219</start>
<end>227</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T When Neutral Turns Negative: Cross-Domain Failure Modes in Hinglish Political Sentiment Analysis
%A Chennuru, Rahul
%A Adebayo, Kolawole John
%Y Afli, Haithem
%Y Bouamor, Houda
%Y Zaghouani, Wajdi
%Y Ghannay, Sahar
%Y Hossain, Shehenaz
%S Proceedings of the 3rd Workshop on Natural Language Processing for Political Sciences (PoliticalNLP 2026)
%D 2026
%8 May
%I ELRA Language Resources Association (ELRA)
%C Palma, Mallorca (Spain)
%F chennuru-adebayo-2026-neutral
%X Sentiment analysis models are increasingly deployed to analyze political discourse, yet strong in-domain performance does not guarantee robustness under domain shift. We study cross-domain generalization in Hinglish (Hindi–English code-mixed) sentiment analysis by evaluating a fine-tuned XLM-RoBERTa classifier, trained on 29,000 general-domain Hinglish sentences, on a curated benchmark of politically oriented Hinglish text. While the model achieves 92.02% accuracy in-domain, performance drops to 71.83% under political domain shift. Error analysis reveals a pronounced directional bias with 48.9% of neutral political statements misclassified as negative, indicating a systematic neutrality-to-negative shift. In addition, 87.5% of incorrect predictions are assigned confidence scores above 95%, pointing to severe miscalibration under distribution shift. We further compare these results against an instruction-tuned large language model (Llama 3.3), which achieves 90.85% zero-shot accuracy and 94.37% accuracy with contextual prompting, while substantially reducing neutrality bias. Our findings indicate the need for domain-aware evaluation, calibration diagnostics, and explicit reporting of failure modes when deploying sentiment models in politically sensitive settings.
%R 10.63317/3evceg5aip7m
%U https://aclanthology.org/2026.politicalnlp-1.24/
%U https://doi.org/10.63317/3evceg5aip7m
%P 219-227
Markdown (Informal)
[When Neutral Turns Negative: Cross-Domain Failure Modes in Hinglish Political Sentiment Analysis](https://aclanthology.org/2026.politicalnlp-1.24/) (Chennuru & Adebayo, PoliticalNLP 2026)
ACL