Unequal Representations: Analyzing Intersectional Biases in Word Embeddings Using Representational Similarity Analysis

Proceedings of the 28th International Conference on Computational Linguistics, pages 1720–1728 Barcelona, Spain (Online), December 8-13, 2020 1720 Unequal Representations: Analyzing Intersectional Biases in Word Embeddings Using Representational Similarity Analysis Michael A. Lepori Department of Computer Science Johns Hopkins University mlepori19@gmail.com Abstract We present a new approach for detecting human-like social biases in word embeddings using rep- resentational similarity analysis. Specifically, we probe contextualized and non-contextualized embeddings for evidence of intersectional biases against Black women. We show that these em- beddings represent Black women as simultaneously less feminine than White women, and less Black than Black men. This finding aligns with intersectionality theory, which argues that mul- tiple identity categories (such as race or sex) layer on top of each other in order to create unique modes of discrimination that are not shared by any individual category. 1 Introduction Word embeddings are ubiquitous components of modern natural language processing systems, and have enabled performance increases on a wide variety of NLP tasks (Devlin et al., 2019; Peters et al., 2018; Pennington et al., 2014). They provide vector representations of words, and are trained on a large cor- pus of text in order to capture the distributional semantics of lexical items. Many studies have shown that these embeddings also encode a variety of dangerous social biases, as these biases are present in naturally-occurring English text (see Mehrabi et al. (2019) for a survey of such work). The present work studies intersectional biases in both contextualized and non-contextualized word embeddings. The theory of intersectionality states that multiple different aspects of a person’s identity often combine to create unique modes of discrimination (Crenshaw, 1989; Crenshaw, 1990). For exam- ple, specific biases against Black women are not necessarily shared by either White women or Black men. Furthermore, discourses on discrimination and public health typically focus on sexism or racism, each implicitly excluding the other (Crenshaw, 1989; Bowleg, 2012). It is further argued that these discourses are focused on the most privileged members of a group (Cren- shaw, 1989). Thus, we might expect that the ostensibly race-neutral category ‘Woman’ is implicitly White, and that the ostensibly non-gendered category ‘Black’ is implicitly male. Using representational similarity analysis (RSA; Kriegeskorte et al. (2008)), we find that both non-contextualized GloVe em- beddings (Pennington et al., 2014) and contextualized BERT embeddings (Devlin et al., 2019) display exactly these biases. We also find that the BERT embeddings of the names of Black women are seman- tically impoverished, compared to the embeddings of the names of White women. Related Work: Many recent studies have sought to understand social biases in NLP systems. Notably, Caliskan et al. (2017) showed that the human biases revealed through the Implicit Association Test (Greenwald et al., 1998) are also present in word embeddings, prompting other studies that seek to detect bias in word embeddings (Kurita et al., 2019; Garg et al., 2018). Other studies have attempted to detect bias in other NLP systems (Rudinger et al., 2018), to automatically discover and measure intersectional biases in word …

Content truncated to 500 words