I am the Digital Publisher for Richmond and it is part of my job to keep abreast of the hot topics and technology trends that are affecting our industry. Right now there is nothing hotter than adaptive learning.
Adaptive learning is part of a wider trending topic: big data. The theory seems to be that as we are able to collect and analyse vast amounts of data taken from all the interactions that we have with technology, we will be able to transform the way we approach all sorts of areas: business, health, sport and obviously learning.
Moneyball is a film by Brad Pitt, based on a true story, where the use of big data transforms the fortunes of a baseball team. The thrust of the argument of the film, and the book it is based on, is that data analysis gives a truer evaluation of a player’s worth than the human knowledge of coaches and team directors.
The real-life team that Moneyball is based on did achieve success but there are also instances where big data has failed spectacularly. One example is Google’s attempt to predict flu outbreaks by analysing search terms – the technology giant’s predictions we consistently incorrect by a factor of 50%.
So what will happen when we apply big data analysis to learning? Will we be able to adapt what content a learner encounters to match their needs? This is the claim of adaptive learning and it certainly sounds very attractive.
In a series of interviews on the eltjam blog, the team from Knewton, the most famous company in the field, gives an explanation of what adaptive learning is and how it can help publishers create content that lends itself to adaptive learning. Knewton have certainly been very adept at promoting themselves and their services but so far we have not seen any concrete evidence of their work in ELT.
My worry and, from the comments in the blog posts from Knewton, it would seem to be a worry that is shared by many ELT educators, is what the content that is created for adaptive learning will look like. Will adaptive learning adapt to ELT or will ELT be adapted to produce the big data that adaptive learning needs to operate?
Some aspects of ELT produce good solid pieces of data, most obviously tests of grammar and vocabulary or the receptive skills of reading and listening. We all know how much easier it is to test these things and how these tests can be quantified into nice percentages. But what about the productive skills of writing and speaking? Again, we know how difficult these are to test, how subjective the results of any speaking tests always are and how many factors need to be considered when trying to grade these skills.
Adaptive learning can only work if it is based on solid, reliable data sets. If we cannot rely on the data produced by speaking and writing activities, because of their subjective nature, maybe adaptive learning will tend to disregard these skills in favour of the areas that do produce reliable data.
Knewton are keen to work with publishers to “create more effective learning materials” but what will these materials look like? Will they been more effective for Knewton, by allowing the company to mine data more effectively? Or will they genuinely be more effective for the learner and help them communicate more effectively? We shall see.