Investigating Gender Bias in Turkish Language Models

arXiv:2404.11726v1 Announce Type: new
Abstract: Language models are trained mostly on Web data, which often contains social stereotypes and biases that the models can inherit. This has potentially negative consequences, as models can amplify these biases in downstream tasks or applications. However, prior research has primarily focused on the English language, especially in the context of gender bias. In particular, grammatically gender-neutral languages such as Turkish are underexplored despite representing different linguistic properties to language models with possibly different effects …

abstract amplify applications arxiv bias biases consequences cs.cl data english english language gender gender bias however language language models negative prior research social stereotypes tasks type web