The Illusion of Personalization: Quantifying Linguistic Diversity in Online Tutoring

Online tutoring, a practice integral to educational adaptability during the post-pandemic era, hinges on effective tutor-student interactions. This study investigates how tutors in virtual settings develop and diversify their language over time, contributing to a growing body of research on interpersonal communication within educational environments. By applying computational language measures to a dataset of online elementary literacy tutoring sessions, we quantify the evolution of linguistic diversity among tutors as they gain experience. Our research examines the development of tutors' linguistic diversity both across and within sessions, and how this diversity relates to the content of their discourse. Analysis of over 2,700 tutoring sessions reveals that tutors exhibit a variety of within-tutor linguistic diversity, with an overall decline over time, suggesting a trend towards more standardized language use. This reduction is more pronounced in off-task interactions compared to on-task instructional content. Furthermore, while between-tutor linguistic diversity remains consistently low, indicating similar communication styles across different tutors, high linguistic diversity within sessions is characterized by longer sentences and more personalized interactions. These findings shine light on incomplete understanding of tutor-student dynamics and of the role personalized communication plays on tutoring success.