Teachers research teacher talk

Scott Thornbury

This paper was presented at the UCLES Diploma TEFLA Conference, London 1996.

A theme that has emerged conspicuously in this conference is the distinction between, on the one hand, "method-based", and, on the other hand, "people-based", teaching (see Bax, this volume). In one guise or another this distinction underlies a frustration frequently expressed at the conference, to wit: that DTEFLA candidates typically "play safe", stick to conventional PPP-type lessons, fail to engage their learners either cognitively or affectively, fail to provide opportunities for authentic language use, fail, in short, to show anything but the most superficial allegiance to a truly communicative approach.

Part of the blame can, of course, be attributed to the demands of the scheme itself - the washback effect of the evaluated lesson, and the security that attaches to having a seamless, unproblematic, but totally constraining lesson plan. The dogged adherence to presentation models of teaching may also be due to the persistent effects of initial training, which, rightly or wrongly, encourages a simplistic cause-effect view of learning and hence a transmission model of teaching. And, maybe, more worryingly, the kind of mechanistic teaching so in evidence is symptomatic of a much more pervasive condition: what I call "new methodism" (in contradistinction to the postmethodism being trumpeted from some quarters - see, for example, Kumaravadivelu 1994). New methodism is the belief - either overt or covert - that language learning is simply learning language, i.e. the conscious mastery of the formal linguistic systems (see Thornbury 1996) - a belief, incidentally, that permeates the design of most current published materials. One could be forgiven for wondering, on the evidence of many assessed DTEFLA lessons, if the Communicative Approach had ever happened at all.

Faced with this methodological intransigence on the part of our DTEFLA candidates, at our centre we have had recource to a programme we call, not without reason, "grammar de-tox". Basically, it is two-pronged. On the one hand we adopt a top-down approach, attempting to subvert the grammar-driven, accuracy-to-fluency model of lesson design, by encouraging experimentation with models that foreground fluency, while still allowing space (albeit a small space) for a focus on form.

The bottom-up approach, which is the subject of this paper, involves the critical analysis, by trainees, of their own teacher talk, on the assumption that it is not so much the presence or absence of so-called "communicative activities" as the quality of interaction in the classroom that determines the degree of authenticity of the language learning process. To borrow a metaphor used by Peter Maingay in his paper: teaching needs to be seen as a kind of conversation; or, as Roger Hunt put it, the teacher needs to develop "the ability to work with the people in the room."

This requires the introduction of a very simple (and by no means exhaustive) taxonomy of teacher-talk features. These include:

  • IRF sequences (initiate-respond-follow up)
  • Feedback on content vs feedback on form
  • Display vs referential questions
  • Wait time
  • Student-initiated questions

IRF sequences, especially where the F consists of feedback on form rather than on content - i.e. the standard "eliciting" technique - have been blamed for constraining the development of authentic discourse in classrooms:

"At times...the IRF structure makes it unattractive and unmotivating for students to participate in classroom interaction, since their responses may be evaluated or examined publicly, rather than accepted and appreciated as part of a joint conversation" (van Lier 1996: 151)

Note that van Lier is wisely cautious ("at times"): the fact is, for many teachers, eliciting is heavily ritualised, and virtually the only way they know how to talk to learners.

With regard to display vs referential questions, Nunan suggests that "it is not inconceivable that the effort involved in answering referential questions prompts a greater effort and depth of processing on the part of the learner" (Nunan 1989: 30). I would go further, and argue that the effort involved in asking referential questions prompts a greater effort and depth of processing on the part of the teacher.

As for wait time, Nunan, again, quotes studies that show that, when teachers are trained to wait 3 or 4 seconds, instead of the customary one, there is not only a decrease in the failure of students to respond, but there is an increase in average length of students' responses. Moreover, the proportion of student initiated questioning increases. All of these adjustments would seem to be worthy objectives in a communicative classroom.

Finally, the class where students ask at least some of the questions - unsolicited - would seem to be a healthy class, not only in terms of the "balance of power", but also on the grounds that, if it is true that language is learned through interaction and the negotiation of meaning, it would seem encumbent on teachers to encourage as much interaction as possible, from all directions, and involving all parties.

Once having introduced trainees to these features, (through, for example, the analysis of short video extracts), and in order to sensistize them to their own use (or abuse) of them, we ask them "to confront their own teaching" (to use Alex Teasdale's expression - see his paper in this volume). Trainees are set an assignment which involves first audio-recording a short section of teacher-fronted classroom interaction in one of their own classes (it often takes two or three attempts before they find a piece that is exploitable). They are then asked to transcribe the segment, and subject it to "discourse analysis" - focusing in particular on the kinds of questions asked - and to evaluate it in terms of its "communicativeness". (As background reading we recommend Nunan 1987, Kumaravadivelu 1993, and Burns 1990).

The transcription and its analysis is then submitted as an assignment and assessed according to both the degree of command of the issues, and the depth of critical reflection it demonstrates.

More importantly, however, (and more elusively) is the extent to which this raised awareness is translated into classroom practice. Entrenched patterns of interaction are neither easily accessible nor painlessly changed. Teachers feel uncomfortable once stripped of their dependence on eliciting, for example. However, in combination with radical adjustments to lesson design, qualitative improvements in teachers' interactions do occur, supporting the claim (and the theme of this conference) that reflection can trigger change.



  • Burns, A. 1990 'Focus on language in the communicative classroom' in Brindley, G. (ed.) The Second Language Curriculum in Action Sydney: NCELTR, 36-58.
  • Kumaravadivelu, B. 1993 'Maximizing learning potential in the communicative classroom" English Language Teaching Journal 47/1, 12-21.
  • Kumaravadivelu, B. 1994 'The postmethod condition: (E)merging strategies for second/foreign language teaching' in TESOL Quarterly 28:1, 27-49.
  • Nunan, D. 1987 'Communicative language teaching: making it work' in English Language Teaching Journal 41/2, 136-145
  • Nunan, D. 1989 Understanding Language Classrooms Hemel Hempstead: Prentice Hall.
  • Thornbury, S. 1996 'Is there life after grammar?' Paper presented at UCLES CTEFLA Conference, February 1996.
  • van Lier, L. 1996 Interaction in the Language Curriculum: Awareness, Autonomy and Authenticity. Harlow: Longman


More articles on language teaching