Very enjoyable discussions on data, software, discourse, power and knowledge were had at the Data in Discourse Analysis conference at the TU Darmstadt on Wednesday 18 February. Drawing partly on analysis with Juliane Jarke on predictive analytics within the DATAFIED project, Felicitas Macgilchrist reflected on how discourse studies’ „object of analysis“ is changing as education is increasingly datafied.
She observes changes to the discourse “about” education, the discourse “in” education, and alerts us to the need to analyse the discourse “encoded into” education. Here’s the updated abstract for the paper:
Datafication, the increased transformation of information about education into digitally manipulable data, is a process occurring in educational practice and in (critical) discourse analytical research on education. In part one of this paper, I explore three ways in which ‘digital data’ have changed the way educational discourse studies perceives its object of analysis. First, ‘data about education’ flow abundantly. Schools are producing more numeric data about their students than ever before. Critical discourse studies have analysed, e.g., PISA, ICILS and other international assessments, the media reports about these assessments, and the networks of ‘experts’ from data science, who are increasingly working alongside government agencies to co-write educational policy. Critical analyses have traced these distributed policy networks, following how discursive elements move across sectors. Second, discourse analyses of classroom practices are investigating ‘data in education’. Studies observe how digital data are transformed into data visualisations, analysing, for instance, the uptake of data visualizations by teachers and students during class time. Third, we can observe ‘data encoded into education’. Here, I see the most substantial change for discourse studies. Algorithms written by specific teams in specific (primarily for-profit) contexts shape the software with which teachers and students engage. Discourse analysts begin to look at code, learning analytics, dashboard design, software documentation. In part two of the paper, a short “worked example” from predictive analytics walks through these three dimensions, reflecting on how data have changed the way discourse studies perceives its object of analysis. The paper ends by suggesting epistemological, political and practical questions for future research.