Scaling up DDL: Challenges of bringing DDL to an online SPOC format

Peter Crosthwaite, University of Queensland

The use of corpora for the purposes of improving academic writing in the form of data-driven learning (DDL) is now thriving, as advances in computer/internet speed, corpus availability, user-friendly corpus query interfaces and data visualisation have come to the fore.  However, the majority of studies in this area have been conducted in face-to-face classroom, teacher-led settings, and generally involving fewer than one hundred participants. This presentation describes the design of a purpose-built short private online course (SPOC) on DDL for error correction in academic writing, due to be rolled out to over 1,000 graduate students.  I discuss the affordances of using a SPOC platform (namely EdX) for self-guided DDL training involving the BAWE corpus within Sketch Engine Open, how DDL has been ‘sold’ to students as a viable method for improving academic writing, and how it has been implemented within the graduate curriculum. I also comment on the challenges involved in using a SPOC for DDL, including deriving worked examples of DDL error correction for SPOC units, the constant need for ‘right’ answers to online DDL activities, how corpus use can be maintained across activities by both L1 and L2 users, providing written corrective feedback within a SPOC context, new strategies for error correction of errors that aren’t typically easy to solve with corpora, and how participation can be meaningfully assessed across 1000s of participants.

NB This course is now open to external parties who want to have a look at it (more extensive usage, like student cohorts using this, can be arranged upon request).  Sign up is via the following link:

https://edge.edx.org/courses/course-v1:UQx+SLATx+2018_S2/about

Recording of talk can be accessed here