Equity Beyond Bias in Language Technologies for Education

Abstract

There is a long record of research on equity in schools. As machine learning researchers begin to study fairness and bias in earnest, language technologies in education have an unusually strong theoretical and applied foundation to build on. Here, we introduce concepts from culturally relevant pedagogy and other frameworks for teaching and learning, and identify future work on equity in NLP. We present case studies in a range of topics like intelligent tutoring systems, computer-assisted language learning, automated essay scoring, and sentiment analysis in classrooms, and provide an actionable agenda for research.

Should You Fine-Tune BERT for Automated Essay Scoring?

“…Meanwhile, in the field of learning analytics, a burgeoning new field of fairness studies are learning how to investigate these issues in algorithmic educational systems (Mayfield et al, 2019;Holstein and Doroudi, 2019). Outside of technology applications but in writing assessment more broadly, fairness is also a rich topic with a history of literature to learn from (Poe and Elliot, 2019).…”

Making Data Science Count In and For Education

“…The development of these types of new methods is especially relevant for data sources that have traditionally been analyzed using qualitative methods: Analyzing audio and visual data, for instance, requires deciding not only what data to model and how to model it, but also to decide what the unit of analysis in audio-visual data is and how to create variables (Bosch et al, 2018;D'Angelo et al, 2019). Further, such methods present challenges regarding the nature of how algorithms process language data, especially of those from marginalized groups--care must be taken in specifying training data and creating algorithms that do not themselves reproduce existing inequities (Mayfield et al, 2019;Zou & Schiebinger, 2018).…”

Making Data Science Count In and For Education

“…The development of these types of new methods is especially relevant for data sources that have traditionally been analyzed using qualitative methods: Analyzing audio and visual data, for instance, requires deciding not only what data to model and how to model it, but also to decide what the unit of analysis in audio-visual data is and how to create variables (Bosch et al, 2018;D'Angelo et al, 2019). Further, such methods present challenges regarding the nature of how algorithms to process language data, especially of those from marginalized groups--care must be taken in specifying training data and creating algorithms that do not themselves reproduce existing inequities (Mayfield et al, 2019;Zou & Schiebinger, 2018).…”