Using Item Response Theory to Measure Gender and Racial Bias of a BERT-based Automated English Speech Assessment System

Using Item Response Theory to Measure Gender and Racial Bias of a BERT-based Automated English Speech Assessment System Alexander Kwako author Yixin Wan author Jieyu Zhao author Kai-Wei Chang author Li Cai author Mark Hansen author 2022-07 text Proceedings of the 17th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2022) Ekaterina Kochmar editor Jill Burstein editor Andrea Horbach editor Ronja Laarmann-Quante editor Nitin Madnani editor Anaïs Tack editor Victoria Yaneva editor Zheng Yuan editor Torsten Zesch editor Association for Computational Linguistics Seattle, Washington conference publication kwako-etal-2022-using 10.18653/v1/2022.bea-1.1 https://aclanthology.org/2022.bea-1.1/ 2022-07 1 7