Using data to ease the grading process - An Educator's Perspective
When I was a junior in high school, our teacher's union declared an order for all the teachers in the district: they were to "work to rule," which meant that they would not spend any time beyond the 40 hours indicated in their contract helping students or grading assignments. This was in lieu of a strike, and I can't blame them for not striking in the freezing cold Minnesota winter.
The effect the "work to rule" had on me as a student opened my eyes to the effort expended by teachers to hand grade assignments, often using their evenings and weekends (without extra compensation) to do so.
Now that I have become an educator, I've personally felt the pain my high school teachers felt. I have begun teaching a SpanishMOOC, and we have over 1,000 students! They submit weekly assignments that are to be hand graded, and I alone simply cannot meet the task of grading these assignments.
Obviously there is no silver bullet, and no proven technology to automatically grade writing assignments as of yet (there is progress in this area, but it still needs work).
I asked volunteers to help me with this task, and we still found it nearly impossible to keep up. So, out of necessity, we developed a process to ease our burden as teachers grading assignments. We classified and categorized our corrections, and made those corrections granular - on a sentence-by-sentence basis. We also developed a self-review cycle so the students could better identify and correct their errors.
Here's how it works:
- Learners follow assignment instructions (e.g. Introduce yourself)
- Learners submit a first draft of their assignment
- Teachers select a sample of assignment submissions
- Teachers grade those submissions and categorize each correction they make (e.g. Express Age)
- We analyze the data generated from the sample set, and find the 10 most common errors
- We create modules that teach students about these common errors and how to avoid them
- We have the learners review their initial drafts, and make corrections to their submissions, then resubmit.
We have found that this substantially reduces the effort we expend on grading assignments. Each assignment, instead of having 10-20 errors (many times repeats of the same mistake), they only have a few.
Another benefit to this approach is the re-usability of these open-ended assignments, both for ourselves and others. We make our content and assignments freely available as Creative Commons, so we include the common errors and the teaching modules that accompany them so that other teachers can skip the data analysis and extraction steps, and benefit from previously collected data.
We want other teachers to help develop assignments too! We’re making our platform available for language teachers to create and grade assignments, and we’ll help analyze the data produced from each one using this same methodology. We hope that with dozens of teachers contributing, we can work together to create a versatile library of assignments and associated principles.
This is just one of the topics we’ll be discussing in our free upcoming MOOC for language teachers, “Blended Teaching of World Languages.” We’ll have discussions and offer technical training on Instreamia, the adaptive platform we are using for SpanishMOOC.
We may not be able to altogether eliminate those late nights of grading from your schedule, but we hope that the platform, methodology, and library of assignments will lessen the burden.
You may also find useful the following video where Ryan and me introduce the next big MOOC: Blended Teaching of World Languages.