From Measurement to Mitigation: Exploring the Transferability of Debiasing Approaches to Gender Bias in Maltese Language Models

Melanie Galea, Claudia Borg, Department of Artificial Intelligence, University of Malta · 2025

Read From Measurement to Mitigation: Exploring the Transferability of Debiasing Approaches to Gender Bias in Maltese Language Models by Melanie Galea, Claudia Borg, Department of Artificial Intelligence, University of Malta (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-03142.txt.

Description

Melanie Galea and Claudia Borg measure gender bias in language models for Maltese in a 2025 study. Maltese is a low-resource language with grammatical gender, so biased model behavior can be easy to miss and hard to fix. Models trained mostly on larger languages carry stereotypes that can harm smaller communities.

The authors first measure how strongly Maltese models show gender bias. They then test whether debiasing methods built for English transfer to Maltese. Such transfer matters because low-resource languages cannot fund separate tool building for every problem. Their findings speak to fairness work beyond the major world languages.

Catalog record

CREATORMelanie Galea, Claudia Borg, Department of Artificial Intelligence, University of Malta
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-03142

Browse the full library · About the Tower of Babel