From Language to Algorithm: Transphobia in Research on Gender Recognition Software
Presenter Title/Affiliation
University of British Columbia
Start Date
23-5-2021 1:30 PM
Event Name
Panel discussion
Panel Number
24
Panel Chair Name
David Peterson
Zoom URL to Join
https://ciis.zoom.us/j/99469567105
Zoom Meeting ID
994 6956 7105
Abstract
Within the North American blogosphere, LGBTQ+ scholars and activists are mounting vocal criticism of the technological drive towards ever more detailed applications of facial recognition and gender identification software (Gault, 2019; Gutierrez, 2019; Hay, 2019; Johnson, 2019; Merlan & Mehrotra, 2019; Samuel, 2019). Much of that critical attention is driven by a small number of scholars in studies of technology and human-computer interaction who challenge the underlying conceptions of gender which inform such software (Hamidi, Scheuerman, & Branham, 2018; Kannabiran & Petersen, 2010; Rode, 2011). Such work has explored the history of algorithmic bias against trans identities (Hicks, 2019), the experiences and strategies trans people use when navigating prescriptive gender norms of technological systems (Ahmed, 2018), and the risks for trans lives that are created by binary and immutable conceptions of gender recognition algorithms (Keyes, 2018). The fact that commercial development of facial recognition software interacts with research practices presents the language employed in research publications as an avenue for examining the presence of trans-inclusive language and trans-competent research design in the development of gender recognition software. Our project investigates the language of facial recognition research publications which mention non-binary gender, gender non-conformity, and gender transition in some way. We are interested in how choices of phrases and citations do or do not draw on trans experience and trans voices, or do or do not link lto scholarship on gender and trans identity (Thieme & Saunders, 2018). For our analysis, we have collected via database searches a corpus of 15 conference and research articles published between 2010 to 2019; we conduct a content analysis with a basic scheme for coding. Given the overwhelming assumptions of binary and immutable conceptions of gender which have been shown to lie at the heart of most technological work on gender recognition (Keyes, 2018), we ask: how do trans, non-binary, or genderfluid identities figure in the research discourse that does mention gender outside an exclusively binary or immutable conception of gender? And how are practices of citation used to characterize the landscape of existing research that is presented as relevant to projects of gender recognition?
Presenter Contact
Katja.Thieme@ubc.edu
mary_ann.saunders@ubc.edu
Laila.Ferreira@ubc.ca
From Language to Algorithm: Transphobia in Research on Gender Recognition Software
Within the North American blogosphere, LGBTQ+ scholars and activists are mounting vocal criticism of the technological drive towards ever more detailed applications of facial recognition and gender identification software (Gault, 2019; Gutierrez, 2019; Hay, 2019; Johnson, 2019; Merlan & Mehrotra, 2019; Samuel, 2019). Much of that critical attention is driven by a small number of scholars in studies of technology and human-computer interaction who challenge the underlying conceptions of gender which inform such software (Hamidi, Scheuerman, & Branham, 2018; Kannabiran & Petersen, 2010; Rode, 2011). Such work has explored the history of algorithmic bias against trans identities (Hicks, 2019), the experiences and strategies trans people use when navigating prescriptive gender norms of technological systems (Ahmed, 2018), and the risks for trans lives that are created by binary and immutable conceptions of gender recognition algorithms (Keyes, 2018). The fact that commercial development of facial recognition software interacts with research practices presents the language employed in research publications as an avenue for examining the presence of trans-inclusive language and trans-competent research design in the development of gender recognition software. Our project investigates the language of facial recognition research publications which mention non-binary gender, gender non-conformity, and gender transition in some way. We are interested in how choices of phrases and citations do or do not draw on trans experience and trans voices, or do or do not link lto scholarship on gender and trans identity (Thieme & Saunders, 2018). For our analysis, we have collected via database searches a corpus of 15 conference and research articles published between 2010 to 2019; we conduct a content analysis with a basic scheme for coding. Given the overwhelming assumptions of binary and immutable conceptions of gender which have been shown to lie at the heart of most technological work on gender recognition (Keyes, 2018), we ask: how do trans, non-binary, or genderfluid identities figure in the research discourse that does mention gender outside an exclusively binary or immutable conception of gender? And how are practices of citation used to characterize the landscape of existing research that is presented as relevant to projects of gender recognition?
https://digitalcommons.ciis.edu/lavlang/2021/sunday/20
