Skip to content

Provide positive comments alongside constructive criticism when responding to student writing

Principle · All principles
Evidence · 21 claims (5 for, 11 mixed, 5 against) · 24 studies (7 review, 6 causal, 4 quant-synthesis, 3 associational, 2 design, 1 qualitative, 1 theoretical), q1–q4 · 5 of 24 report an effect size · 15 claims rest on one study

Conditional relationship

This page owns what a teacher's response to a student's draft says about what works in it, beside the critique. For a student writer (an adolescent or adult, a novice or intermediate writer in the genre, including second-language writers) who will revise the draft, a response that names specific strengths of this draft (what the writing does and why it works for its reader: "this example makes your claim concrete", not "good job" or "you are a gifted writer") alongside a small number of prioritised critical comments is expected to keep the writer reading and acting on the critique (motivation-affect, persistence into revision) and so to produce a better revised draft (skill-performance on a writing-quality judgement), compared with critique alone, with praise in place of critique, or with generic or person-directed praise. The goal is a complex-skill (composing a text); a secondary target is the writer's self-confidence and motivation to write (attitude-motivation), which the earlier page named.

The relationship is conditional on four things. The positive comment must carry information about the work: feedback helps on average and helps more the more information it carries, and feedback that turns attention to the person helps least and can lower performance. Praise must not replace critique: a strength comment frames the critique, and a response with praise and no usable critique withholds the feedback the writer needs. Praise for ability is a risk when failure may follow: children praised for being smart did worse after a later failure than children praised for effort. The writer must have a chance to act on the response (another draft), or neither strength nor critique can change the text. It is proposed for classroom, online-instructor-led and online-self-paced settings over days-weeks or term-plus, with teacher, peer or automated comments on drafts.

No claim in the wiki tests this relationship: none compares writing feedback with and without specific positive comments, or measures a revised draft against the balance of praise and critique. The earlier page rests on one practitioner article (Shvidko 2020), which has no claim page here. What the claims test are neighbouring parts: what any feedback must carry to help (two meta-analyses), what praise for ability does to children after failure (one set of experiments and one review), whether verbal praise undermines free-choice engagement as tangible rewards do (two competing meta-analyses), and how automated feedback distributes praise and critique (two studies of AI-generated feedback, one of them on student essays). Two brief "affirmation" findings that the earlier page cited bear on the identity-threat reason it gave, and both limit it.

How the converted neighbours sit beside this page, so a designer can go to the one that holds each decision. Process-based Writing holds the draft–respond–revise cycle and the reader stance in responding; this page holds one move inside the response, naming strengths. Wise Feedback Across Difference holds critique framed with a high standard and an assurance that the learner can reach it, for learners who may doubt the giver's intent; its "positive comments" part is this page's move, and its framing should be added when that doubt is likely. Strengths-Based Approach holds how a learner's demonstrated capabilities are named and built on across tasks; a strength comment on a draft is one such naming, at the grain of one text. Feedback Loops holds whether a comment reaches a next attempt that is checked again. The structured peer review pattern holds the peer routine in which students give each other praise and critique.

Default design, while the relationship is untested

The earlier page's guidance, kept as a concrete default a teacher or designer can act on and revise. Each step is a design proposal; its evidence status is stated beside it.

  1. Read for what works before marking what does not. On a first read, mark two or three places where the draft does something well for its reader. Untested proposal; from the earlier page ("acknowledge positive aspects of student work and highlight the strengths").
  2. Make each strength comment specific to the text. Name the move and its effect ("Your second paragraph answers the objection a sceptical reader would raise"), not the person ("You're a natural writer") and not a bare tick. Short marginal remarks ("Well said", "Good point"), which the earlier page reports can build motivation, are fine as additions beside a passage, not as the whole response. Basis: feedback carrying more information helps more, and feedback aimed at the self helps least (Feedback improves learning by a medium amount on average, but its effect varies widely with the information it carries, and a substantial share of feedback effects are negative [+S]; Feedback Most Effective At Task And Process Levels [+S]); untested for strength comments on writing.
  3. Praise the choice or the work, never the writer's ability. Comment on decisions the writer made (a strategy, a revision, effort spent on evidence) rather than talent. Basis: Praise for intelligence, rather than for effort, after success undermines children's motivation and performance after a later failure [-S] (fifth graders on reasoning problems, not writing); carried to older writers untested.
  4. Give the critique, prioritised: two or three comments, the most important first. Choose what the next draft most needs (usually ideas, organisation or evidence on an early draft), and say what to do, not only what is wrong. Basis for keeping critique in: LLM feedback for students marked by race, language, disability, gender, or motivation shows positive feedback bias and feedback withholding bias — more praise and less substantive critique than for unmarked counterparts [-M] (AI-generated feedback on 600 essays; shows the risk, not its effect on learning). The number of comments is an untested proposal.
  5. Do not set a praise-to-critique ratio by rule, and do not make praise automatic. Match strength comments to what the draft actually does; a draft with little that works gets one honest strength and focused critique, not invented praise. Basis: The agent's praise was indiscriminate affirmation: validation followed about half of participant turns regardless of what the participant said [-M] (a reflection agent in a randomized trial, not writing feedback; it shows what content-blind praise looks like, not its effect). The earlier page notes that the balance "may not always be easy to achieve".
  6. End with the next step. Close the response with what to do in the next draft and, where the writer is likely to doubt your intent, an explicit statement of the standard and that you think they can meet it (modelled in Wise Feedback Across Difference). Untested proposal here.
  7. Check what the writer did with it. On the next draft, look first at whether the prioritised comments were acted on, and name any improvement as the first strength comment ("The new topic sentence now tells the reader where this is going"). Untested proposal; links progress to the strength comments.
  8. Plan the arc across drafts and the term, not one response. A default arc, all quantities untested proposals:
  9. First draft of a new genre (early in a term): 5–10 minutes per draft; two or three strength comments on ideas and structure, two or three critical comments on the biggest issue, one next step. No line editing.
  10. Second draft: about 5 minutes; open with one comment naming what improved since the first draft, then one or two critical comments, which may now reach paragraphs and sentences.
  11. Final draft: a short summary of what works and one target for the next piece of writing; grade separately.
  12. Across the term: move the source of the strength comments from the teacher to the writer. From the second or third assignment, ask writers to mark what they think works and what they want help with before submitting; respond to their choices, agreeing, adding or correcting. Peer responses using PQP (Praise-Question-Polish) can supply the first round once writers have seen several teacher responses.
  13. Moving on: a writer whose own strength notes match yours on two consecutive drafts, and who acts on critique without prompting, needs fewer teacher strength comments; keep them specific when you do give them.

When a condition is not met: if there is no further draft, give the strength and the one change for the next piece of writing rather than a list of faults; if the writer only reads the grade, return comments before the grade or without it (see the table below).

Fitting the design to a situation

The three facts that most change the decision:

  • Whether the writer will revise this draft. Comments on a draft that will be revised can be acted on; comments on a final text are read, if at all, as a justification of the grade. If unstated, ask: will the writer produce another version of this text, and when?
  • How the writer is likely to read critique. A confident writer may need few strength comments; a writer with low confidence, a second-language writer, or one with reason to doubt the teacher's intent may stop reading at the first criticism. If unstated, ask: how has this writer responded to critique before, and do they trust the person giving it?
  • Who writes the comments, and with how much time. A teacher with ten minutes per draft, a peer, and an AI tool produce different balances of praise and critique. If unstated, ask: who will respond to each draft, and how long do they have?
If the brief says… Then change… Basis
Writers are novice in the genre or have low confidence as writers Open with two specific strength comments before any critique; keep critique to one or two priorities with a concrete next step; name the improvement on the next draft first untested proposal; from the earlier page (motivation and confidence of student writers). The information requirement is Feedback Most Effective At Task And Process Levels [+S], not tested on writing confidence
Writers are advanced or professional, and ask for critique One strength comment where it marks something to keep; spend the response on critique at the level they ask for untested proposal; the earlier page's balance is meant for developing writers
Learners are child (primary age) Praise the strategy or effort used ("you added a detail that shows how she felt"), never ability ("you're a great writer"); keep comments spoken or very short Praise for intelligence, rather than for effort, after success undermines children's motivation and performance after a later failure [-S] (fifth graders, reasoning problems); Praise focused on global intelligence has negative effects on future learning behavior compared with praise about effort [-M] (the same experiments, reported by a practice guide)
Writers are second-language learners Separate comments on ideas from language comments; praise a language form used correctly when it was the target; limit language critique to one or two error types untested proposal; the earlier page names language learners among its writers
The writer belongs to a group subject to negative stereotypes about ability, or has reason to doubt the teacher's intent Keep the full critique; add the statement of the standard and the assurance that the writer can reach it; do not soften critique into praise Wise Feedback Across Difference (untested there too); the risk of withholding critique is shown for AI feedback in LLM feedback for students marked by race, language, disability, gender, or motivation shows positive feedback bias and feedback withholding bias — more praise and less substantive critique than for unmarked counterparts [-M]
Comments come from an AI tool (online, online-self-paced) Before using it, read a sample of its comments on weak and strong drafts and on drafts by different students: check that weak drafts still get specific strengths, that strong drafts still get critique, and that comments do not change with the student's described identity or motivation; do not tell the tool a student's group or motivation LLM feedback sentiment tracks assigned scores: low scores come with more criticism, high scores with more praise [~M]; Motivation prompts shift feedback tone: unmotivated students receive enthusiastic, collaborative encouragement while motivated students receive direct, higher-level critique [~M] (both about the comments produced, not their effect on writers)
An AI tool is set to open with encouraging, learner-centred feedback before elaborated feedback Do not assume the encouraging layer helps learning; give the specific critique in the first response and check revised work Non-layered AI feedback produced significantly higher post-test performance than layered AI feedback after controlling for pre-test [-M] (one online lesson, undergraduates, not writing; layering and encouragement were not separated)
One teacher for a large class, or scarce responder time (a busy term, volunteer tutors, one short check-in) Per draft: one specific strength, one priority change, one question, written in under 3 minutes; or a whole-class note on common strengths and one common problem, with individual comments only on the priority. Return comments before the grade Students report teacher feedback on writing is insufficient, and they value scores over feedback [~M] (one interview study of Chinese college English students, who reported teachers lacked time and that they cared more about scores); the minutes are an untested proposal
Peers give the feedback Use a structure that requires a specific strength before critique, such as PQP (Praise-Question-Polish), and model one strength comment and one weak one ("nice") first untested proposal; the peer routine is held by structured peer review
The brief names its own routine, rubric or criteria for responding (its own comment template, rubric bands or categories) Keep it. Write strength comments in the rubric's own criteria and words, and add only what the evidence says matters: that each strength names something in the text, and that critique is not dropped untested proposal (wave 5 and 6 review lessons)
The writing is private or sensitive (a journal, a personal narrative) Respond only if and as the writer chooses; comment on the writing's craft (what makes a moment vivid), not the experience; no critique of content the writer did not offer for critique untested proposal
Written praise from this responder could be misread (a facilitator of a different sex, a setting with approval rules) Check the setting's rules before using written personal notes; keep strength comments about the work, in the shared document or rubric, or give them aloud School-level constraints prevented two of nine sites from using the four-part written praise strategy [~M] (a youth programme evaluation, not writing instruction)
Single session only, or a fixed short time (for example 40 minutes) One product revised once: about 15 minutes drafting, 5 minutes of response (one strength, one priority) from teacher or peer, 15 minutes revising, 5 minutes for the writer to name what changed. Adds no test or survey untested proposal
High stakes: a graded final text with no revision Give the grade with a short note of what works and one target for the next piece; do not spend time on extensive critique the writer cannot use untested proposal; Students report teacher feedback on writing is insufficient, and they value scores over feedback [~M]
Adults writing for work (reports, case notes), or a product for a real audience Name strengths in the terms the real reader would use ("a manager can act on this summary"); keep the whole text whole rather than splitting it into exercises; critique against what the audience needs untested proposal
Mixed modes: some learners remote, some in the room, or phone-only Use comments anchored in the shared document so everyone sees them in place; short audio or text comments for phone-only learners, one strength and one priority each untested proposal

Seven of the seventeen rows cite a claim; none of those claims tests strength comments on student writing. Two carry findings about children's reasoning problems to young writers, three describe what AI tools produce rather than what learners gain, one is an interview study, and one is a programme evaluation outside writing. The rest are untested proposals.

Observation, state and explanation

What can be observed: the comments given (how many strengths, whether each names something in the text, how many critical comments and which were marked priority), whether the writer opened or read them, what changed between drafts, and what the writer says about the response. "Discouraged", "defensive" and "open to feedback" are inferred states.

Response Plausible explanations Observation that could change the interpretation
The revised draft changes nothing the critique asked for The comments were not read (grade only); the critique was unclear; the writer read the critique as a judgement of them and disengaged; no time to revise Ask the writer to say, in their own words, what the main comment asked; check whether the grade was returned with the comments.
The writer fixes surface errors only Surface errors are what they think revision means; the priority comment was buried among many; the higher-order critique was too vague to act on Return a draft with only the priority comment and a next step, and see what changes.
The writer keeps the praised feature and repeats it everywhere The strength comment was read as a rule; it named a feature but not why it worked here Ask why the feature worked in that place, and whether it would work in another.
A confident writer disregards strength comments They already know the strengths; the comments were generic Compare the writer's own list of strengths with yours.
A writer says the feedback was harsh although strengths were given Strengths were generic and the critique specific; the writer doubts the responder's intent; the volume of critique swamped the rest Count specific versus generic comments, and ask the writer which comment they remember first.

These rows are untested proposals as diagnostics. Asking a writer about the response is itself a response, so record each probe.

Evidence and qualifications

No claim in the wiki compares writing feedback with and without specific positive comments, or measures revised drafts or writing motivation against the mix of praise and critique. The claims below test neighbouring parts of the relationship.

Writers, tasks and outcomes differ across these claims (children's reasoning problems, laboratory tasks, AI comments on essays, an online lesson), so do not rank them. None measures the quality of a revised draft.

The earlier page cited two findings for its reason, that feedback can threaten a writer's identity and that affirmation buffers the threat. Both limit rather than support it:

The evidence by question

Further evidence, not yet read against this model

Claims found while converting this page that bear on it at a distance. None tests positive comments on writing.

Objective and learner-valued goal

The teacher's objective is usually a better revised text and, over a term, a writer who can judge their own drafts. Elicit separately what the writer wants from a response: a grade, reassurance, to know whether the piece works for its reader, or to get a job application or report accepted. A writer who wants only the grade may not read comments at all; one who wants reassurance may read only the strengths. Agreement and divergence are both observations: ask the writer what they want feedback on before responding, and respond to that as well as to your priority.

For example, a second-language university student may want to know whether their English "sounds natural", while the teacher's priority for this draft is the argument's structure. Name a strength in the structure, give the structural critique, and add one comment on the language point the writer asked about, so the writer's question is answered without displacing the priority.

What would revise this model?

The expectation should weaken if comparisons of the same critique with and without specific strength comments, on comparable writers, show no difference in revision quality or in whether writers act on the critique. If generic praise ("good job") and specific strength comments produce the same revisions and the same motivation to write, the information requirement here is a preference, not a condition. If strength comments reduce the critique writers act on (by signalling that the draft is good enough), the default should give critique first. If writers who doubt the responder's intent respond to the standard-and-assurance framing but not to strength comments, the move belongs to Wise Feedback Across Difference rather than here.

Writers' reported confidence, whether they read the comments, what changed in the next draft, and the quality of later independent writing are separate outcomes. The present evidence establishes no number of strength comments, no ratio of praise to critique, and no effect of either on writing.

Examples

Key Sources

  • Shvidko, Elena. (2020). Taking into Account Interpersonal Aspects of Teacher Feedback: Principles of Responding to Student Writing. Journal on Empowering Teaching Excellence, Vol. 4, Iss. 2, Article 7. https://doi.org/10.26077/936a-72f7
Page metadata
Field Value
type principle
id provide-positive-comments-on-student-writing
title Provide positive comments alongside constructive criticism when responding to student writing
description For a student writer who will revise a draft, a response that names specific strengths of the text alongside a few prioritised critical comments is expected to keep the writer acting on the critique and to improve the revision, more than critique alone or generic or ability praise; no claim in the wiki tests this, and the nearest evidence (feedback meta-analyses, praise experiments with children, AI feedback audits) bounds it.
status review
canonical True
generated by claude/unspecified on 2026-10-15
trust unverified — no one has confirmed this page's content
sources 1 — shvidko-2020

Frontmatter is the page's machine-readable record — see the Schema & Guide.