Scott Thornbury

iTDi Research Issue – Scott

On Research and Taking it Forward

Scott Thornbury

I guess I’m a big picture kind of person. I like to see (and report on) major shifts and trends, and generalize these to the widest range of contexts. So I have some problems with the issue of research – not the idea, so much as the reality. It seems to me that it is difficult to reconcile the tension between wanting to prove something worthwhile, on the one hand, and, on the other, needing to train one’s lens as narrowly as possible, in order to eliminate extraneous ‘noise’, and to deliver results that will be even remotely credible.

Years ago, Peter Strevens warned that ‘perhaps the biggest single piece of self-discipline required of the individual who undertakes a programme of research is to limit his [sic] subject, to cut it down and prune it and cut it down again until it really is a single project capable of being completed in a reasonable time’. And he adds, ‘the almost universal tendency is to take on too large a subject and to become so embroiled that it is never completed, or else it becomes dangerously superficial’ (1968:27).

‘Dangerously superficial’ is my middle name! I would love, for example to demonstrate that you can learn a language in classrooms simply by using it. But to do that I will have to (1) define ‘using it’ in ways that are measurable; (2) conduct research with a wide range of learners in terms of age, nationality, first language, motivation, level, etc – and (3) for each combination have a control group that is evenly matched with the experimental classes in every particular, including the teacher. Impossible, frankly.

So, I am reduced to conducting a much smaller-scale study: for example, following just one group of learners over a relatively short period of time, and comparing the results of a post-test with a pre-test. Whatever results I get will be suggestive at best, unlikely to be significant, and easily refutable by anybody who doesn’t want to believe them.

Of course, there’s the ‘grain of sand’ argument, i.e. that a lot of small-scale studies like mine can add up to one big conclusion. This is the point of meta-analyses of the type that Norris and Ortega (2000) did, where they crunched together the results of lots of micro-studies to generate some fairly robust conclusions about the role of instructed learning. But to do a meta-analysis properly, you have to verify the tiniest details of every study, a tedious and possibly inconclusive process.

All this assumes that the point of research is to prove something. But, of course, research can be an end in itself. Like Cavafy’s ‘Ithaca’, it may be that the destination is less important than the journey. What I learn along the way, either by changing some aspect of my practice, or by scrutinizing it more closely, can only serve me in good stead. In these terms, research is simply one stage of the larger reflective cycle, without which professional development is at risk of atrophying.

However, there are some issues I would really like to see investigated, not just for the sake of the journey, but for any results, however tentative, that the research might yield. Many of them relate to the way that language emerges, particularly through interaction. And many of them overflow into one another. For example,

  • What do learners learn from each other (e.g. in pair and group work) and how durable is this learning, compared to other possible sources?
  • Is there evidence that memorized whole phrases (or chunks) are re-analyzed into their constituents, and thereby ‘release’ their grammar, at some point?  If so, which kinds of phrases, and under what conditions?
  • By the same token, what are the conditions by which isolated elements (words, phrases) are chunked into larger units that are produced and interpreted holistically? That is to say, can grammar emerge out of a kind of bricolage process?
  • Is there any correlation between phase shifts (i.e. sudden leaps forward) in one system, e.g. grammar or pronunciation, and phase shifts in another (e.g. vocabulary)? And is the correlation a causal one, e.g. does the acquisition of a critical mass of vocabulary trigger changes in grammar? If so, can we put a number on it?
  • Just what do learners ‘appropriate’ (if anything) from their teachers in ‘scaffolded’ conversations? Again, what does it take before these externally-sourced items become internally-regulated? I.e. how does ownership occur, and how can the teacher facilitate the process?
  • How much incidental learning occurs in a language class? That is, how many words or expressions do learners pick up that weren’t necessarily targeted by the teacher? How durable is this learning, and what can be done to optimize it?

Any takers?



Norris, J.M. and Ortega, L. (2000) ‘Effectiveness of L2 instruction: a research synthesis and quantitative meta-analysis’, Language Learning, 50, 417-528.

Strevens, P. (1968) ‘Linguistics and research in modern language teaching’, in Jalling, H. (ed.) Modern Language Teaching, London: Oxford University Press.

Published by

Scott Thornbury

Scott is a teacher and teacher educator, with over 30 years' experience in English language teaching. He is currently Associate Professor of English Language Studies at the New School in New York, teaching on an on-line MATESOL program. His previous experience includes teaching and teacher training in Egypt, UK, Spain, and his native New Zealand. Scott’s writing credits include several award-winning books for teachers on language and methodology including The A-Z of ELT, How to Teach Grammar and Teaching Unplugged. He is series editor for the Cambridge Handbooks for Teachers (CUP) and was also the co-founder of the dogme ELT group, whose archived website, called Teaching Unplugged, can be found below. Scott currently leads a fascinating community at the popular and thought-provoking blog, A-Z of ELT blog. Scott is lead author in the iTDi Teacher Development program as well as being iTDi's Academic Director.

5 thoughts on “iTDi Research Issue – Scott”

  1. Thank you for the post Scott, I loved (and am experiencing!) the fact that you raise the issue of having to chisel things down into imposed boundaries and the often unspoken reality of time and there just needing to be so much of it to produce the quality of work deemed ‘worth’ reading.

    The issue then for me is who does it get read by and who does it directly or even ultimately benefit… where does it all end up? The most transforming moment in my approach to research was when I was told to stop blogging because I was now doing a doctorate…I haven’t. Thank goodness 🙂

    1. Thanks for your swift response, Divya. Stop blogging indeed! I consider blogging my primary vector of research. I pose questions, collect data (from the comments) and test hypotheses!

  2. Scott, thank you so much for the post!

    Like the great Socrates, you don’t give us answers but pose a series of questions to spark the curiosity and thought in us, and to encourage us to research, analyze and reflect! Sometimes such questioning can be worth tons of manuals and guidebooks. So thanks a lot for your questions: they gave me a lot of food for thought.

    The two questions (one on how much students learn from each other and the other on incidental learning) are the most relevant to what I’m doing in my university classroom. Due to the lack of time, I have to exploit pair and group work a lot. Moreover, for the same reason I have to rely solely on the peer teaching when it comes to vocabulary. This has made me think on the issues you touched upon in your post: I’ve started asking myself how effective and durable this technique is. The answers to these questions will undoubtably have an immediate impact on my teaching: I will either stop leaving it all to my students and take on a more active role, or I will perfect the existing technique and help students become even more autonomous and confident learners.

  3. Thanks, Alexandra, for your comment. I suspect that you and I are not alone in (almost desperately!) seeking validation of the benefits of peer collaboration. Fortunately the research paradigm that is providing some support is the domain of sociocultural learning (i.e. Vygotskian-influenced cognitive science). The work that Merrill Swain (in Canada) and her colleagues have been doing recently is worth looking at here.

    1. Scott, thank you very much for mentioning Merrill Swain and her research. I haven’t heard of her before, but now I’ve looked through the titles of some of her works, and I see that they are most relevant to what I’ve been concerned with. No doubt, Merrill’s ideas will help me to come up with some simple techniques to measure the efficiency of peer teaching and incidental learning. Thank you.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.