The Red Pen on Trial

AA teacher's red pen is among the most familiar objects in education and among the least examined. A workload survey published in 2021 found that secondary teachers in the country studied here spent an average of 5.2 hours a week writing on pupils' work, more than they spent on any activity other than teaching itself. Yet what that time buys has been contested for decades. Feedback, in the definition most researchers now use, is information a learner receives about work already completed, offered in order to change the work that follows; a mark, which ranks, and praise, which reassures, are on this definition not feedback at all. The first controlled studies, conducted between 1958 and 1971, were designed to establish whether pupils needed to be told that their answers were wrong. The second wave, gathering pace after 2015, has mostly been commissioned by ministries worried about teachers leaving the profession rather than about learning itself. Whether classroom experiments running for two or three years can license a national marking policy is what now divides the field.

BThandiwe Lindholm, an educational psychologist at the Halstead Centre for Learning Research, has spent seven years on a trial conducted in 41 secondary schools between 2016 and 2019, in which 3,400 pupils aged fourteen each wrote four extended essays. Classes were assigned at random to three conditions: a mark alone, written comments alone, or both together. The experiment was commissioned to settle whether removing marks would depress attainment, and on that question its result was undramatic: pupils given comments alone finished two marks higher on a hundred-point scale than those given a mark alone, a gap Lindholm herself calls too small to justify any policy. What the trial is now cited for is something it was never built to measure. Among pupils receiving comments without a mark, the proportion who rewrote at least one paragraph before handing the work back rose from 31 per cent to 68 per cent, and the gap between conditions widened during the second year rather than closing. Lindholm regards this as the finding that has best survived scrutiny, precisely because nobody had gone looking for it.

CThe result carries a boundary that Lindholm states far more often than those who quote her repeat it. Every essay collected in the trial was a piece of analytical writing in history, because that was the only subject in which the participating schools shared a common marking rubric. The trial therefore says nothing about mathematics or the practical subjects, which are precisely the lessons whose marking loads are complained of most. One design feature does travel further than the rest: teachers in the comments-only classes were given ten minutes for each essay and told to write a single instruction rather than a list of corrections, and Lindholm attributes the revision result partly to the fact that a pupil was asked to carry out one instruction rather than weigh several. She has written twice that the figures should not be described as evidence that marks are harmful, since every pupil in the study still received a grade at the end of each term.

DZoltan Lindstrom, a specialist in school organisation at the Redcliffe School of Education, does not contest the revision finding and says so plainly. His objection is that an experiment and a timetable are different objects. Applied to every essay a pupil writes across every subject, the ten-minute rule would take a full-time teacher about nine hours a week, well above the figure the workload survey had recorded, and no trial anywhere has yet run for longer than three years. That horizon is short enough, he argues, that teachers keep to a protocol they would quietly abandon in a fourth year, once a novel instruction had hardened into a routine chore. Put to him directly, he concedes two things: his costing assumes that no existing marking is given up, an assumption he agrees inflates the figure, and his claim about drift cannot itself be tested until some school holds to the method for a decade.

EMinistries have not waited for the argument to end. Since 2019 at least twenty-two education authorities have issued new guidance on marking, and the announcements travel a good deal further than the results. A policy requiring one written instruction on every essay and a policy forbidding written comments altogether are reported under the same phrase, and their headline figures are then set side by side as though they had measured one thing. Neither Lindholm nor Lindstrom defends the practice. Both name it the most misleading habit in the field, and both have argued that any figure quoted from a classroom trial ought to be published together with the subject and the age group it came from. It is the rare point on which the two agree without qualification, and neither of them expects the newspapers to oblige.

FWhat is emerging is a division of labour rather than a verdict. Where the work is a draft that can be rewritten, such as an essay or a laboratory report, a written comment has no substitute, because a pupil cannot act on a number. Where the aim is narrower, such as correcting a spelling error that half the class repeats, a minute of whole-class instruction still delivers more for each minute spent, and neither researcher disputes it. The question no trial has yet answered is what a comment does to a pupil who knows that a machine wrote it: in every study so far the author was a teacher the pupil met each week, and trust, unlike accuracy, is earned in the corridor rather than on the page.