Saved in:
Bibliographic Details
Main Authors: Das, Amit, Hasan, Md. Najib, Sarkar, Souvika, Zhang, Zheng, Jamshidi, Fatemeh, Bhattacharya, Tathagata, Raychawdhury, Nilanjana, Feng, Dongji, Jain, Vinija, Chadha, Aman
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.22720
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917091312926720
author Das, Amit
Hasan, Md. Najib
Sarkar, Souvika
Zhang, Zheng
Jamshidi, Fatemeh
Bhattacharya, Tathagata
Raychawdhury, Nilanjana
Feng, Dongji
Jain, Vinija
Chadha, Aman
author_facet Das, Amit
Hasan, Md. Najib
Sarkar, Souvika
Zhang, Zheng
Jamshidi, Fatemeh
Bhattacharya, Tathagata
Raychawdhury, Nilanjana
Feng, Dongji
Jain, Vinija
Chadha, Aman
contents Large Language Models (LLMs) have demonstrated remarkable proficiency in generating text that closely resemble human writing. However, they often generate factually incorrect statements, a problem typically referred to as 'hallucination'. Addressing hallucination is crucial for enhancing the reliability and effectiveness of LLMs. While much research has focused on hallucinations in English, our study extends this investigation to conversational data in three languages: Hindi, Farsi, and Mandarin. We offer a comprehensive analysis of a dataset to examine both factual and linguistic errors in these languages for GPT-3.5, GPT-4o, Llama-3.1, Gemma-2.0, DeepSeek-R1 and Qwen-3. We found that LLMs produce very few hallucinated responses in Mandarin but generate a significantly higher number of hallucinations in Hindi and Farsi.
format Preprint
id arxiv_https___arxiv_org_abs_2507_22720
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Investigating Hallucination in Conversations for Low Resource Languages
Das, Amit
Hasan, Md. Najib
Sarkar, Souvika
Zhang, Zheng
Jamshidi, Fatemeh
Bhattacharya, Tathagata
Raychawdhury, Nilanjana
Feng, Dongji
Jain, Vinija
Chadha, Aman
Computation and Language
Large Language Models (LLMs) have demonstrated remarkable proficiency in generating text that closely resemble human writing. However, they often generate factually incorrect statements, a problem typically referred to as 'hallucination'. Addressing hallucination is crucial for enhancing the reliability and effectiveness of LLMs. While much research has focused on hallucinations in English, our study extends this investigation to conversational data in three languages: Hindi, Farsi, and Mandarin. We offer a comprehensive analysis of a dataset to examine both factual and linguistic errors in these languages for GPT-3.5, GPT-4o, Llama-3.1, Gemma-2.0, DeepSeek-R1 and Qwen-3. We found that LLMs produce very few hallucinated responses in Mandarin but generate a significantly higher number of hallucinations in Hindi and Farsi.
title Investigating Hallucination in Conversations for Low Resource Languages
topic Computation and Language
url https://arxiv.org/abs/2507.22720