Authors: Ye Chen; Bei Yu; Yihan Yu
Addresses: Department of Educational Technology and Foundations, College of Education, University of West Georgia, Carrollton, GA 30116, USA ' School of Information Studies, Syracuse University, Syracuse, NY 13244, USA ' Department of Human Centered Design and Engineering, University of Washington, Seattle, WA 98195, USA
Abstract: Reflection journaling is a common practice in teacher education. However, analysing large amounts of textual reflections presents challenges. Automatic analysis is needed so that teacher educators could quickly uncover valuable patterns and provide adaptive, real-time support. This study proposed a text-mining method to discover the important themes and patterns in preservice teachers' reflection journals. We also examined the potential text features through which the quality of reflection could be assessed. A total of 367 journals from 80 preservice teachers were analysed. The results showed that our text-mining method was able to accurately identify the weekly teaching focus and the themes in which participants had long-standing interest. We found that when participants engaged in higher-level of reflection, their journals achieved higher topic relevance to weekly teaching focus and they tended to write longer reflections. Based on the text-mining results, we further developed prediction models to automate the assessment of written reflections.
Keywords: text mining; topic modelling; reflection; reflection engagement; journal writing; teacher education.
International Journal of Innovation in Education, 2021 Vol.7 No.2, pp.122 - 143
Received: 02 Sep 2020
Accepted: 06 Apr 2021
Published online: 28 Sep 2021 *