Translation Without Talk

Every activity in the routine exercises one channel - reading and copying - and none exercises production.

• Input is always pre-digested. The teacher's line-by-line translation removes the need to guess meaning from context, so students never practise the strategy that real reading requires.
• There is no pushed output. Only 5 of 44 volunteer because speaking has never been a required, low-risk routine; only a public performance. The other 39 have correctly learned that silence is safer.
• The guide is a rational response, not laziness. If answers are always copied from the board, students infer that the answer exists outside their heads; when a free-writing task appears, the guide is the only model they have.
• Twenty red marks teach avoidance. When correction means every error is displayed at once, the cheapest defence is to write less, or to copy text that contains no errors.

Diagnosis: the class has decoding and translation, but no information gap and no audience. Fix those two and the rest follows.

Task: 'Find the difference' (picture A / picture B).

• Prepare two simple line drawings of the same village scene with eight differences (a well present/absent, two cows/three cows). No projector is needed; two photocopied sets are enough. If photocopying is unavailable, draw two versions on card and pass them round in plastic sleeves.
• Students work in fixed pairs: the front-facing student gets A, the one in front of them turns around to get B. They may not show the picture; they must talk. Target language is pre-taught and written on the board: There is / There isn't / Is there a ...? / How many ...?
• Model once with a strong student for 90 seconds so the class hears what success sounds like.
• Timing: 2 minutes teacher model, 3 minutes paired talk round 1, swap roles, 3 minutes round 2, 6 minutes feedback collecting five differences from five different pairs, 3 minutes writing three sentences about the differences.

Keeping 44 on task: use a visible signal (raised hand, then all hands) to stop; circulate along four fixed routes; nominate from the silent majority, not the five volunteers. Success indicator is not accuracy but the number of pairs still talking English at minute eight.

Task redesign. Replace the single open prompt with a two-part item that requires local, unguessable content:

1. Draw a simple plan of the nearest crossroads and label four things (3 marks).
2. Write five sentences saying what you can see there and one change you would like (5 marks).

A guidebook cannot contain the plan of this crossroads, so the copied paragraph loses its value. Add one sentence in the first person about a named place, and copying becomes near-impossible without effort greater than writing.

Correction policy for 44 scripts.

• Use selective marking: choose one error type per piece of work (this fortnight, verb forms only) and mark only that. Everything else is ignored on paper but noted in a class error tally.
• Return a class error board, not 44 individual corrections: the five most common errors, with one corrected example each, taught in ten minutes.
• Use two-minute oral feedback with six students per week on rotation instead of full written correction; that reaches all 44 in seven weeks and costs less time than marking everything.
• For the March item, choose a short choral poem or a dialogue with strong rhythm - success in performance rewards fluency, which the marking scheme has so far never rewarded.