A learner receives an English essay marked 78/100. Before reading anything else, the number settles the emotional question: good enough, disappointing, or better than last time. The margin comments arrive after that verdict and compete for attention, while a Taiwanese senior-high or university learner who needs the score for a transcript still has to receive a grade. Teachers can change the order. Give descriptive feedback and revision time first. Reveal the grade after the learner has used the comments. A two-day delay can create room for both without withholding a final record.
The number ends the search
A grade compresses a complex performance into one position on a scale, which is useful for certification, reporting, and course decisions while giving the learner little direction for the next draft. A 78 can hide an accurate argument with weak paragraph links, frequent article errors, and two sentences copied too closely from the source. The same 78 can describe a different paper with clean grammar and thin evidence.
A hypothetical cover sheet reads: 78/100. Clear claim. Article errors. See paragraph 2.
When a grade and comments share the page, the number is easier to process. It answers “How did I do?” in half a second. Written feedback asks for slower work: locate the passage, infer the teacher's standard, compare options, and plan a revision. Students under time pressure may glance at the prose, compare scores with a friend, and put the paper away.
That sequence can protect slower work.
Comments and grades compete
Ruth Butler's 1988 experiment assigned 12 classes of fifth- and sixth-grade pupils to comments, numerical grades, or grades plus comments. The published abstract reports results for 132 randomly selected pupils at higher and lower achievement levels. Interest and performance were highest after comments. Grades and grades plus comments produced similar patterns overall, though high-achieving pupils responded differently on some measures when they expected further grades. The children completed one convergent and one divergent task across two sessions, which limits direct comparison with an adult English course held over a term.
Grades plus comments deserve attention. Placing advice beside a score failed to combine both benefits. The evaluative signal could dominate task-focused information.
Anastasiya Lipnevich and Jeffrey Smith tested a related question with 464 students in an introductory psychology course, where participants revised a single essay examination under combinations of detailed feedback, a tentative grade, praise, and an attributed source. Their 2009 experiment found that detailed, task-specific feedback was strongly associated with higher revision scores. Descriptive feedback worked best when it arrived without grades or praise. A tentative grade generally reduced performance, although praise changed that effect in some conditions. Whether students believed the feedback came from an instructor or a computer made little difference in this task.
These controlled studies answer narrow questions. Butler studied children completing convergent and divergent tasks. Lipnevich and Smith studied one psychology essay. Their project did not follow second-language learning across a term. Neither experiment shows that every learner ignores comments beside a grade, or that schools should stop reporting achievement. The findings support a smaller design choice: separate the moment for improvement from the moment for evaluation when revision is still possible.
Build two moments
Consider a hypothetical GEPT writing class. On Tuesday, the teacher returns a draft with two coded language patterns and one comment about the reader's difficulty following the second paragraph. The score box is covered. The learner has 17 minutes to label each problem, choose one sentence to rewrite, and add a transition that states the logical relationship. A partner reads only the revised paragraph and marks the sentence where the argument becomes clear. On Thursday, after the revision is submitted, the teacher releases the rubric scores and a short note showing which descriptor changed. The grade still exists. It now closes a completed feedback cycle instead of interrupting one.
| First return | After revision |
|---|---|
| Two error patterns to track | Rubric score |
| One reader problem to solve | Change from the prior draft |
| A concrete revision request | Record for the course |
This design also keeps feedback small enough to use.
Marking every error produces a dense page and an exhausted teacher. Choc Education's article on following fewer writing errors across drafts explains why a narrow target gives both learner and teacher a better record of change.
Make the learner act
Comments gain value through the next action. “Good organization” gives pleasant information and no decision. “Move sentence 6 above the example, then add the reason the example supports your claim” produces a visible revision. “Articles” written in the margin names a category. Asking the learner to find all four noun phrases with missing determiners turns the category into retrieval and editing. The learner should leave a changed sentence, a new paragraph order, or another result that the teacher can inspect without guessing what happened during revision.
Keep the first-return routine short:
- Read all comments without a pen for 2 minutes.
- Rewrite each comment as an action in the learner's own words.
- Choose one action that changes meaning and one that changes language accuracy.
- Submit the changed lines with a brief note identifying what was altered.
The teacher can then check the response instead of rereading the whole paper. If a learner misinterprets the comment, that misunderstanding surfaces before the grade fixes attention on rank. We prefer delayed grades when a real revision follows within a few days. The preference has boundaries. A final external examination offers no second draft. A placement decision may need an immediate score. Some learners also require prompt reporting for an application deadline. In those cases, a separated sequence may be impractical, and a short conference can direct attention to one usable comment beside the score.
Praise needs precision too
Praise can pull attention toward identity in the same way a grade pulls it toward rank. “You are a talented writer” offers no map for the next paragraph. A precise observation works differently: “Your example in paragraph 3 gives a date, actor, and consequence, so the claim can be checked.” The learner can repeat those features elsewhere without having to defend a label about talent. For the next marked assignment, cover the score with a removable note. Write two prompts on the cover: identify the first sentence to change, then state what the reader will understand after that change. Let the learner answer and revise before lifting the note.