Socially Aware Bias Measurements for Hindi Language Representations

Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1041 - 1052 July 10-15, 2022 ©2022 Association for Computational Linguistics Socially Aware Bias Measurements for Hindi Language Representations Vijit Malik1∗ Sunipa Dev2 Akihiro Nishi2 Nanyun Peng2 Kai-Wei Chang2 1Indian Institute of Technology Kanpur (IIT-K) 2University of California, Los Angeles {vijitvm}@iitk.ac.in {akihironishi}@ucla.edu {sunipa, violetpeng, kwchang}@cs.ucla.edu Abstract Trigger warning: This paper contains exam- ples of stereotypes and other harms that could be offensive and triggering to individuals. Language representations are efficient tools used across NLP applications, but they are strife with encoded societal biases. These bi- ases are studied extensively, but with a primary focus on English language representations and biases common in the context of Western soci- ety. In this work, we investigate biases present in Hindi language representations with focuses on caste and religion-associated biases. We demonstrate how biases are unique to specific language representations based on the history and culture of the region they are widely spo- ken in, and how the same societal bias (such as binary gender-associated biases) is encoded by different words and text spans across lan- guages. The discoveries of our work highlight the necessity of culture awareness and linguis- tic artifacts when modeling language represen- tations, in order to better understand the en- coded biases. 1 Introduction Language models and representations (Pennington et al., 2014; Bojanowski et al., 2017; Devlin et al., 2019) are commonly used across the world in a variety of applications including machine trans- lation (Kunchukuttan et al., 2017), information retrieval (Rao and Devi, 2018), chatbots (Bhag- wat et al., 2019), sentiment classification (Kumar et al., 2019) and more. However, it is also known that these representations capture and propagate societal biases including gender (Bolukbasi et al., 2016), race (Caliskan et al., 2017), (Manzini et al., 2019), and nationality (Dev and Phillips, 2019) related stereotypes. This bias is present across rep- resentations for different languages. Each language reflects the culture and history of regions they are used popularly in, and as we go from one language ∗Work done while interning at UCLA. to another, the notion of bias, and the types of soci- etal biases change accordingly. This key difference however, is not reflected in the effort made towards detecting, identifying, and mitigating biases in lan- guage representations, with the majority of efforts predominantly in English language and in the con- text of Western society (Bolukbasi et al., 2016; Nangia et al., 2020). Some recent work tackle the challenges of societal biases in language rep- resentations coming from various cultures and lan- guages such as Arabic, French, Spanish, Chinese, and German (Lauscher et al., 2020; Ch´avez Mulsa and Spanakis, 2020; Kurpicz-Briki, 2020; Zhou et al., 2019; Zhao et al., 2020; Liang et al., 2020). Additionally, Ghosh et al. (2021) and Sambasivan et al. (2020) explore biases and algorithmic fair- ness concerning non-western contexts in machine learning and Pujari et al. (2019) focus on fairness in language technologies for Indic society …

Content truncated to 500 words