Clover has been fairly quiet for the last few months and mostly that is on account of her keeper being wedged down a rabbit hole without the benefit of keys or shrinking potions. No amount of irritable hoof-tapping from Clover could redeem the situation. I simply had to sit it out until the grip of circumstance released me and I popped out, fractionally slimmer, into the light and sweetness of the paddock.
As luck would have it, it’s a lovely new paddock and I am now happily engaged as a Research Associate with the Centre for Learning and Research in Higher Education at the University of Auckland. In October, along with my colleagues on our Ako Aotearoa funded project, Building an Evidence-Base for Teaching and Learning Design Using Learning Analytics Data, we’ll be heading round New Zealand for a series of workshops, seminars and conversations about learning analytics. We are visiting most of the main centres: Auckland, Hamilton, Wellington, Christchurch and ending in Dunedin. We’re looking forward to lots of good discussion, debate and opportunities to share practice.
My focus will be a workshop designed to introduce teachers to convenient and simple approaches to analysing large volumes of student generated text; specifically, responses to short-answer questions. To this end, I’ve created a Jupyter notebook that demonstrates an analysis pipeline. It begins with text responses and ends with a wordtree. Wordtrees were first described by Martin Wattenberg and Fernanda B. Viégas and are a lovely way to represent a keyword in context (KWIC). KWIC or concordances, are very familiar to linguists and language teachers but arguably less familiar to educators in general. They can be incredibly useful if you are trying to understand how people are using a particular word or phrase in a given context. As an example, here’s a wordtree with a focus on the modal verb need as used by tertiary teachers responding to a question about ethics and learning analytics.
Participants who are handy with coding may find the notebook useful but there are lots of less technical options for achieving the same end and this is where the workshop will be firmly focussed. To me, the exciting thing is that text analysis has the potential to take us well beyond the click-counting, engagement proxies and predictive modelling that is the preoccupation of much learning analytics research. Analysing student responses along with the language that we use as teachers gets to the heart of teaching and learning conversations – and to the heart of student learning.
Find out more about the Learning Analytics Roadshow