How we got here: Short-scale change in identity labels for trans, cis, and non-binary people in the 2000s

By Lal Zimman, Will Hayworth. 2020

Though understudied in research on language variation and change, the lexicon is a crucial domain for sociopolitical transformations of language. This paper presents a corpus-based sociolinguistic analysis of changes in terms for transgender, cisgender, and non-binary individuals in four online communities on the social media blogging site, LiveJournal.com – one for trans women, one for trans men, one for non-binary people, and another for transgender people in general – that were popular in the 2000s. Using innovative corpus methods that utilize general purpose cloud computing tools, we focus on changes in the popularity of labels for trans, cis, and non-binary people, the factors that impact the variable use of these terms, and what kinds of differences can be observed across the four LiveJournal communities of practice studied. It thereby contributes both to the study of language and identity in trans and queer communities and to the development of methods for studying large datasets of technologically-mediated communication.

Subfields: AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In AI Ethics, Fairness & Society

1. Douwe Kiela
2. Melanie Kambadur
3. Eric Michael Smith
4. Eleonora Presani
5. Adina Williams

Citations in the corpus

• “I’m sorry to hear that”: Finding New Biases in Language Models with a Holistic Descriptor Dataset (2022, methodology): “, 2018), and since people references are prone to change over time (Smith, 1992; Galinsky et al., 2003; Haller et al., 2006; Zimman and Hayworth, 2020), we have taken inspiration from calls to make model evaluation more dynamic (Kiela et al.”
• "I'm sorry to hear that": finding bias in language models with a holistic descriptor dataset (2022, methodology): “, 2018), and since people references are prone to change over time (Smith, 1992; Galinsky et al., 2003; Haller et al., 2006; Zimman and Hayworth, 2020), we have taken inspiration from calls to make model evaluation more dynamic (Kiela et al.”
• Perturbation Augmentation for Fairer NLP (2022, methodology): “We encouraged annotators to use these conventions, but they are unlikely to be uniformly applied, as human group references are prone to change over time (Smith, 1992; Galinsky et al., 2003; Haller et al., 2006; Zimman and Hayworth, 2020).”