Welcome to the Modern World of Pronouns: Identity-Inclusive Natural Language Processing beyond Gender

Proceedings of the 29th International Conference on Computational Linguistics, pages 1221–1232 October 12–17, 2022. 1221 Welcome to the Modern World of Pronouns: Identity-Inclusive Natural Language Processing beyond Gender Anne Lauscher MilaNLP Università Luigi Bocconi Milan, Italy anne.lauscher@unibocconi.it Archie Crowley Linguistics University of South Carolina Columbia, SC, USA acrowley@sc.edu Dirk Hovy MilaNLP Università Luigi Bocconi Milan, Italy dirk.hovy@unibocconi.it Abstract Trigger warning: This paper contains examples that contain transphobic language. The world of pronouns is changing – from a closed word class with few members to an open set of terms to reflect identities. However, Nat- ural Language Processing (NLP) barely reflects this linguistic shift, resulting in the possible ex- clusion of non-binary users, even though recent work outlined the harms of gender-exclusive language technology. The current modeling of 3rd person pronouns is particularly problem- atic. It largely ignores various phenomena like neopronouns, i.e., novel pronoun sets that are not (yet) widely established. This omission contributes to the discrimination of marginal- ized and underrepresented groups, e.g., non- binary individuals. It thus prevents gender equality, one of the UN’s sustainable devel- opment goals (goal 5). Further, other identity- expressions beyond gender are ignored by cur- rent NLP technology. This paper provides an overview of 3rd person pronoun issues for NLP. Based on our observations and ethical consid- erations, we define a series of five desiderata for modeling pronouns in language technology, which we validate through a survey. We eval- uate existing and novel modeling approaches w.r.t. these desiderata qualitatively and quan- tify the impact of a more discrimination-free approach on an established benchmark dataset. 1 Introduction Pronouns are an essential component of many lan- guages and often one of the most frequent word classes. Accordingly, NLP has long studied tasks related to them, e.g., pronoun resolution (e.g., Hobbs, 1978). Simplistically, they can be defined as “a word (such as I, he, she, you, it, we, or they) that is used instead of a noun or noun phrase”.1 1Essential definition provided by the Mirriam Webster Online Dictionary at https://www.merriam-webste r.com/dictionary/pronoun Nom. Acc. Poss. Poss. Reflexive (dep.) (indep.) Gendered Pronouns he him his his himself she her her hers herself Gender-Neutral Pronouns they them their theirs themself Neopronouns thon thon thons thons thonself e em es ems emself xe xem xyr xyrs xemself ey em eir eirs emself e em eir eirs emself ze zir zir zirs zirself ... Nounself Pronouns star star stars stars starself vam vamp vamps vamps vampself ... Emojiself Pronouns s s self s s self ... 0Numberself Pronouns 0 0s 0s 0self 31 31 1 3s 1 3s 1 3self ... Nameself Pronouns John John Johns Johns Johnself ... Table 1: Non-exhaustive overview of phenomena related to third-person pronoun usage in English. Linguistic studies have pointed out the complex- ity of pronouns, though (e.g., Postal et al., 1969; McKay, 1993). Pronouns can carry demographic information – in English, for example, information about the number of referents and a single refer- ent’s (grammatical) gender.2 Pronouns can con- vey even more information …

Content truncated to 500 words