Part 7: data driven learning

So what is data driven learning? Well, data driven learning involves the use in the classroom of computer-generated concordances to get students to explore regularities or patterning in the target language, and the development of activities and exercises based on concordance output. So data driven learning works by getting the students to discover things about language, almost taking on the role of researcher themselves, rather than having the facts spoon fed to them or having to learn things by rote. So an example of data driven learning would be something like the sort of language exercise that Granger and Tribble have used, one of which I’ve put on the slide. In this example, we’ve got an exercise for teaching the word “accept”.

We’ve got a series of concordances from both native and learner speakers of English and some questions. What grammatical structures appear to follow “accept”? And do any of the graphical forms only appear in the learner examples? Are the students using an acceptable form? So the students would look down these concordances and perhaps come to the conclusion of the learner speaker ‘s example six and eight are not acceptable uses. So they could say that “accept” tends to be followed by words like “that”, “the”, and “what” and then a noun phrase. But accept shouldn’t be followed by the word “to”, for example. So learner corpora are useful in letting us focus on potentially problematic uses of language learning.

And in particular, we can look at different types of errors or we can focus on over or underused linguistic phenomena. However, in order to establish norms about the target language, one also often uses learner corpora in conjunction with an L1 corpus, a reference corpus of some sort of L1 productions, native speaker productions.