@inproceedings{patel-etal-2026-romanized,
title = "From {R}omanized to {D}evanagari: Enhancing {N}epali Sentiment Analysis with {N}epali{X}lit",
author = "Patel, Suraj and
Dhami, Kashish Kumari and
Sherpa, Norden and
Khadka, Supriya",
editor = "Sarveswaran, Kengatharaiyer and
Vaidya, Ashwini",
booktitle = "Proceedings of the Second workshop on Challenges in Processing {S}outh {A}sian Languages ({CH}i{PSAL}2026)",
month = may,
year = "2026",
address = "Palma de Mallorca, Spain",
publisher = "ELRA Language Resources Association (ELRA)",
url = "https://aclanthology.org/2026.chipsal-1.12/",
doi = "10.63317/5j8kkeb2myf6",
pages = "115--126",
abstract = "Romanized Nepali is the dominant medium of social media communication in Nepal, yet most multilingual NLP models are trained on Devanagari, creating a noticeable drop in performance in informal settings. To address this script mismatch, we develop NepaliXlit, a transliteration model fine-tuned from IndicXlit to better handle the phonetic variability of Romanized Nepali. Trained on 2,943 informal word pairs and evaluated on 736 held-out pairs, NepaliXlit improves transliteration accuracy by 8{\%} and reduces character error rate by 11{\%}. We use sentiment analysis as a testbed to understand whether transliteration actually helps downstream NLP tasks. We curate over 6,500 Romanized social media comments and construct a balanced subset of 1,518 manually annotated instances. Baseline experiments show that multilingual encoder models struggle with Romanized input; however, transliterating text into Devanagari using NepaliXlit consistently improves sentiment classification accuracy with mBERT and MuRIL. Comparative evaluation against large language models (LLMs) further reveals that generative models such as Gemini and GPT variants exhibit strong cross-script generalization and outperform encoder-based baselines. Our results indicate that adaptive transliteration enhances conventional multilingual models, while modern LLMs offer a better alternative for multi-script, low-resource settings."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="patel-etal-2026-romanized">
<titleInfo>
<title>From Romanized to Devanagari: Enhancing Nepali Sentiment Analysis with NepaliXlit</title>
</titleInfo>
<name type="personal">
<namePart type="given">Suraj</namePart>
<namePart type="family">Patel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kashish</namePart>
<namePart type="given">Kumari</namePart>
<namePart type="family">Dhami</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Norden</namePart>
<namePart type="family">Sherpa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Supriya</namePart>
<namePart type="family">Khadka</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 Second workshop on Challenges in Processing South Asian Languages (CHiPSAL2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kengatharaiyer</namePart>
<namePart type="family">Sarveswaran</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ashwini</namePart>
<namePart type="family">Vaidya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA Language Resources Association (ELRA)</publisher>
<place>
<placeTerm type="text">Palma de Mallorca, Spain</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Romanized Nepali is the dominant medium of social media communication in Nepal, yet most multilingual NLP models are trained on Devanagari, creating a noticeable drop in performance in informal settings. To address this script mismatch, we develop NepaliXlit, a transliteration model fine-tuned from IndicXlit to better handle the phonetic variability of Romanized Nepali. Trained on 2,943 informal word pairs and evaluated on 736 held-out pairs, NepaliXlit improves transliteration accuracy by 8% and reduces character error rate by 11%. We use sentiment analysis as a testbed to understand whether transliteration actually helps downstream NLP tasks. We curate over 6,500 Romanized social media comments and construct a balanced subset of 1,518 manually annotated instances. Baseline experiments show that multilingual encoder models struggle with Romanized input; however, transliterating text into Devanagari using NepaliXlit consistently improves sentiment classification accuracy with mBERT and MuRIL. Comparative evaluation against large language models (LLMs) further reveals that generative models such as Gemini and GPT variants exhibit strong cross-script generalization and outperform encoder-based baselines. Our results indicate that adaptive transliteration enhances conventional multilingual models, while modern LLMs offer a better alternative for multi-script, low-resource settings.</abstract>
<identifier type="citekey">patel-etal-2026-romanized</identifier>
<identifier type="doi">10.63317/5j8kkeb2myf6</identifier>
<location>
<url>https://aclanthology.org/2026.chipsal-1.12/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>115</start>
<end>126</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Romanized to Devanagari: Enhancing Nepali Sentiment Analysis with NepaliXlit
%A Patel, Suraj
%A Dhami, Kashish Kumari
%A Sherpa, Norden
%A Khadka, Supriya
%Y Sarveswaran, Kengatharaiyer
%Y Vaidya, Ashwini
%S Proceedings of the Second workshop on Challenges in Processing South Asian Languages (CHiPSAL2026)
%D 2026
%8 May
%I ELRA Language Resources Association (ELRA)
%C Palma de Mallorca, Spain
%F patel-etal-2026-romanized
%X Romanized Nepali is the dominant medium of social media communication in Nepal, yet most multilingual NLP models are trained on Devanagari, creating a noticeable drop in performance in informal settings. To address this script mismatch, we develop NepaliXlit, a transliteration model fine-tuned from IndicXlit to better handle the phonetic variability of Romanized Nepali. Trained on 2,943 informal word pairs and evaluated on 736 held-out pairs, NepaliXlit improves transliteration accuracy by 8% and reduces character error rate by 11%. We use sentiment analysis as a testbed to understand whether transliteration actually helps downstream NLP tasks. We curate over 6,500 Romanized social media comments and construct a balanced subset of 1,518 manually annotated instances. Baseline experiments show that multilingual encoder models struggle with Romanized input; however, transliterating text into Devanagari using NepaliXlit consistently improves sentiment classification accuracy with mBERT and MuRIL. Comparative evaluation against large language models (LLMs) further reveals that generative models such as Gemini and GPT variants exhibit strong cross-script generalization and outperform encoder-based baselines. Our results indicate that adaptive transliteration enhances conventional multilingual models, while modern LLMs offer a better alternative for multi-script, low-resource settings.
%R 10.63317/5j8kkeb2myf6
%U https://aclanthology.org/2026.chipsal-1.12/
%U https://doi.org/10.63317/5j8kkeb2myf6
%P 115-126
Markdown (Informal)
[From Romanized to Devanagari: Enhancing Nepali Sentiment Analysis with NepaliXlit](https://aclanthology.org/2026.chipsal-1.12/) (Patel et al., CHiPSAL 2026)
ACL