Speaking your mind?

Speaking in an exam

[image from playbill.com)

“I don’t mind talking to you or my classmates in English; in fact, I quite enjoy it. But it’s different when I have to have a conversation with a stranger whose job is to look for mistakes in what I say and how I say it – the examiner is not there to support me, but to judge me. And they’re not interested in talking to me,  they just want to do their job, pass or fail me, and then go home!”

This is how a learner preparing for a public exam described his attitude to the speaking test. And he’s actually quite right. His words capture two salient characteristics of speaking tests, which both teachers and learners need to remember:

  • A speaking test is not a normal communicative interaction. The main purpose is neither to exchange information, nor to exchange opinions, nor to express feelings. The main purpose is for the learner to demonstrate what they know and what they can do in English and for the examiner to judge the learner’s performance.
  • The examiner’s job is, indeed, to make decisions about the learner’s level based on the evidence the learner provides. In most EFL examination contexts, the examiner is actually discouraged from speaking naturally and participating naturally in a conversation with the learner; instead, they are asked to follow a script in the interests of standardization and fairness.

An unnatural situation

The ensuing unnaturalness of the situation is not an accident, but rather an inherent necessity, one of the necessary evils, it seems, of exams. Speaking examiners, who are mostly teachers themselves, learn to check their teaching personality at the examination room door and put on the hat of the impartial, but occasionally robotic, examiner. In a similar way, learners may also have to learn that it is in their best interest to adopt a speaking style for the examination room which may be very different from what is the norm in natural conversation.

Research into natural conversation (for example, Leech 2000, Quaglio and Biber 2006, Miller 2006) has shown that it displays characteristics like the following:

  • at least half of the utterances are fragments, incomplete clauses and parenthetics
  • more than 60% of the complete clauses used have a very simple subject – verb – object structure
  • more than 70% of the verbs are in the present tense
  • there are quite a few structures (including non-defining relative clauses, participial clauses and infinitive clauses) which are extremely rare
  • lexical density and  lexical diversity are both very low, which means that a limited number of words are used, rare words are avoided and many of the words are repeated again and again
  • cohesive devices are limited to very simple conjunctions like and and but, with only occasional use of conjuncts.

And yet, to prove that they have reached an advanced level, learners taking a speaking exam are expected to demonstrate that they can construct full, grammatical sentences, that they can use a broad range of grammatical structures, that they have an appropriate range of vocabulary at their disposal and that they can employ a variety of cohesive devices to help them connect their ideas and signal what logical relationships obtain between different ideas. In other words, the kind of speaking performance that will earn learners good marks in a speaking test seems to display none of the characteristics of natural conversation; worse still, speaking tests seem to require that learners should use a register that is more typical of written than of spoken English. 

What should we teach them, then?

That doesn’t necessarily mean that teachers should be teaching learners how to speak unnaturally. In my view, we should instead make sure that learners understand what the requirements of the exam are and make clear when we are practising  exam skills rather than practising the language. Rather than pretending that what learners have to do in the exam is simply “be themselves” and interact the way they would in “real life” (which, unsurprisingly, is a shorthand term in the ELT world for “life outside the classroom”), I think we should clarify that an exam is most certainly not a normal communicative context, that it has its own rules and its own conventions, that it is, at the end of the day, a spoken genre different  from almost all the others, which learners can master, given the right guidance.

Sheldon or Penny?

Such guidance would include explicit references to the assessment criteria used in the exam and activities focusing on evaluating spoken performance in relation to those criteria. This, however, could be introduced as a game: listening to normal conversations and deciding whether they would be  good enough for an exam as well as listening to less normal conversations and deciding whether those would be good enough for an exam. Which of the two characters, for example, in the video below would you say interacts normally? And which one demonstrates a good range of structures and vocabulary that examiners would appreciate?

Learners in the know

Once learners realise that they need to play a role in the examination room, for which they need to develop a different manner of speaking, one that shows off their language knowledge and skills, it will be possible to continue teaching them real English as well as preparing them for the exam without feeling as though we were all taking part in an absurd performance of an absurd play.  All we have to do is keep the two separated: this is real life, that is exam survival!


Leech, G. 2000. Grammars of spoken English: new outcomes of corpus‐oriented research. Language Learning 50 (4):675‐724
Quagluio P. and D. Biber 2006. The Grammar of Conversation. In B.Aarts and A. McMahon (Eds.) The Handbook of English Linguistics. Oxford:  Blackwell.