Twelve minutes a script, and nobody reads the red ink

Feedback problems:

• Cognitive overload. Correcting every error means the learner receives more information than they can act on; the red page communicates "you are wrong in many places" and nothing more specific. Selective correction is not laziness, it is the condition for uptake.
• No uptake loop. If the essay is never rewritten, the feedback never becomes learning. My colleague's improvement is probably not caused by having fewer errors marked but by the rewrite cycle.
• Correction types are mixed and unprioritised. Spelling, article use and paragraphing sitting side by side implies they are equally important. They are not, at B1, for this task.
• Focus on error only. Nothing in my marking tells the learner what they did well enough to do again.

Marking problems:

• The holistic 6–7 cluster is a symptom of an unusable scale, not of a uniform class. A single impressionistic number cannot distinguish content, range, accuracy and organisation, so it drifts with mood and time of day.
• No anchors. Without benchmark scripts or descriptors written as observable features, my judgement is not defensible to a learner, a parent or a moderator.

Implication: I need an analytic rubric with three or four criteria (task achievement/communicative effect, organisation, range, accuracy), four bands with concrete descriptors, and I need to mark one criterion at a time across a batch, moderate with my colleague on a sample of three scripts, and re-mark a couple blind later to check stability.

The design principle: fewer comments, a compulsory revision, and the learner doing the work.

1. Before writing (in class, 10 min). Give the task with a short genre model and a checklist of three things the rubric rewards. Learners write one paragraph in class so I can catch problems early rather than at submission.

2. While marking (5 min per script, not 12). Mark against three focus points drawn from the class's current needs - say, past-tense consistency, linking words, and paragraphing - using a code, not full rewrites. Two ticks for things done well. No comment on anything outside the focus points except where meaning breaks down.

3. Return (15 min of lesson time). Learners do not get the script until they have done a self-check against the same checklist. Then they find and correct their own coded errors in a different colour. This is where most of the learning happens and it is why my colleague's drafts improve.

4. Rewrite (homework). Second draft compulsory, submitted with the first. I do not re-mark everything; I check only the focus points and whether the coded errors are fixed.

5. Record. A one-line note per learner in a mark book: focus point achieved / not yet. Over a term this shows progress far better than a running column of 6s and 7s.

I would not fight either expectation head-on; I would meet them and add to them.

On "correct everything":

• Explain the rationale once, briefly and in terms of their goal: "If I mark everything, you fix nothing. Three things, you fix three things." Show them the second draft of a previous essay as evidence.
• Meet the need behind the request - reassurance about progress - by giving the rubric criteria and showing which band they are in and what moves them up. Learners who can see the ladder ask for red ink less.
• Keep the holistic mark available but make it the summary of the analytic scores, not a substitute for them.
• Offer the detail they want in a different place: a short one-to-one slot, or a personal target list, rather than on the page.

On marks out of 10:

• Produce the analytic score first, then convert to a total out of 10 using a published weighting, so the number is a by-product of defensible judgements rather than an impression.
• Moderate: mark three scripts alongside my colleague, compare, agree adjustments, keep the anchor scripts.
• Check consistency by re-marking a sample blind a week later; if a script moves by more than half a band, my descriptors are too vague and need rewriting.
• Keep the descriptors behavioural, so I can say to a learner "this paragraph has a topic sentence and two supporting points" rather than "this feels like a 6".

What I would not do is keep the 12-minute red-ink script because it feels thorough. Thoroughness the learner never acts on is not feedback; it is documentation of my effort.