Unit 3 - mini-lecture
I hope by now you are convinced that the data-driven approach is useful for language learning. But of course not every approach is appropriate for every context. I've hinted that certain applications of concordancer-based lesson creation may be more appropriate for more advanced learners. It's also probably worthwhile to consider other aspects of the learning situation when planning a lesson using data-driven learning.
Standard considerations for lesson planning, such as the age of the students, size of class, relative importance of technology in the learning environment, etc. certainly still apply. In addition, there is evidence that the data-driven approach's effectiveness is influenced by the aspect of language being learned. The grammatical structures, types of vocabulary and level of discourse being learned should all probably be carefully considered before any data-driven approach is decided on.
Wisdom about how effective data-driven learning (DDL) is in these various contexts exists in two forms: research results, and anecdotal evidence from educators who have used a DDL approach in designing learning. In a study by Tian(2005) the DDL approach was shown to be more effective than a "traditional" approach in for students learning word usage distinctive features of types of words but not for grammar. Additionally, the study showed that the data "favored students with a higher proficiency" slightly.
Other studies have produced similar results, supporting the intuition that using concordancer output may be more appropriate for a) higher proficiency students and b) vocabulary acquisition rather than grammar. But DDL does allow for a more "holistic" method of acquiring vocabulary. Students are able to learn more than they would from a contrived vocabulary lesson, and much more than they could from a dictionary entry. Learning words in authentic grammatical contexts surely at least reinforces existing grammatical proficiency. Exposure to diverse examples of authentic use is probably a more effective way to fully learn words than reading lists, no matter how good a learner is at memorization.
It may also be possible to have some control over the how effective DDL is for different proficiency levels if you have some control over the corpus being used, since the difficulty of the language presented is a function of what is in the corpus. Of course, one of the benefits of using corpora is the vast amount of authentic language. But, as we have seen, concordances can be made from any set of texts. So, by perhaps sacrificing some authenticity, you can produce concordancer output from texts at any level difficulty.
References
- Tian, Shiauping (2005). Data-Driven Learning: Do Learning Tasks and Proficiency Make a Difference? Proceedings of the 9th Conference of the Pan-Pacific Association of Applied Linguistics, 360-394.
- Boulton, Alex (2009). Testing the limits of data-driven learning: language proficiency and training. ReCALL, 21 : 37-54 Cambridge University Press doi:10.1017/S0958344009000068
- St. John, E.(2001). A case for using a parallel corpus and concordancer for beginners of a foreign language. Language Learning & Technology, 5(3), 185-203.
