The Annotated Review: A Co-Review Protocol for Teaching Peer Review as Writing Instruction

Peer review is one of the few scholarly genres that most researchers are expected to perform without formal training. Graduate students and postdocs often learn it by watching a mentor’s final comments, if they see them at all. The result is a hidden curriculum: trainees absorb standards of evidence, tone, and confidentiality unevenly, and mentors lose a chance to make scholarly judgment visible.

A co-review protocol addresses that gap. In its simplest form, a trainee drafts reviewer comments and a mentor annotates the reasoning behind each one. The manuscript remains confidential; the teaching happens in the margin. The goal is not to produce a junior reviewer who imitates a mentor’s voice, but to make the criteria of peer review explicit enough to be practiced, questioned, and improved. Co-review belongs in science communication because it treats review as a form of scholarly writing—one that can be taught.

From Shielded Trainee to Supported Reviewer

Many labs protect trainees from peer review until they are deemed ready. That caution is understandable. Review is confidential, consequential, and easy to do badly. But shielding trainees also delays the moment when they must learn to separate a weak claim from a strong one, to distinguish a major concern from a stylistic preference, and to write criticism that is direct without being dismissive.

Co-review offers a middle path. The trainee is not asked to review alone, and the mentor is not asked to surrender editorial control. The trainee drafts; the mentor annotates; the final review is submitted under the mentor’s name or as a disclosed co-review, according to the journal’s policy. In U.S. graduate programs, where professional development plans increasingly include conference presentations and paper drafts, co-review can occupy a similar place: a supervised scholarly activity with visible learning goals.

The Three-Stage Workflow: Read, Rank, Rewrite

The protocol works best when it is divided into three stages, each with a short written product. The stages can be completed over two or three meetings, or asynchronously with tracked comments.

Stage 1: Read

The trainee reads the manuscript without drafting comments. They write a 150-word summary of the manuscript’s central claim, the evidence offered, and the main limitation they see. The mentor checks the summary for comprehension before any critique begins. This step prevents a common failure mode in novice review: a long list of objections that misses what the paper is actually trying to do.

Stage 2: Rank

The trainee lists concerns and ranks them by severity and evidential basis. For each concern, they note the relevant section or figure and state what evidence would resolve it. The mentor annotates the ranking, explaining why one concern is major and another is minor. A concern about an unsupported causal claim may outrank a missing citation; a concern about a figure’s axis label may be minor but still worth mentioning. The trainee learns that review is not a tally of imperfections but a judgment about which imperfections affect the paper’s claims.

Stage 3: Rewrite

The trainee drafts reviewer comments. The mentor annotates each comment with the reasoning behind it and, where needed, a revision that models a more precise or less inflammatory tone. The trainee rewrites. The final review should make sense to an editor who has not been part of the mentoring conversation. This stage is where writing instruction happens most directly: the trainee sees not only what to say but why one phrasing asks for clarification while another assigns blame.

What Annotated Comments Look Like

Good co-review comments teach evidence standards without turning the review into a lecture. Consider a comment that begins, The statistics are wrong. A mentor might annotate it as follows: name the specific claim, identify the mismatch between methods and analysis, and request clarification rather than pronouncing failure. The revised comment could read: The claim in Figure 3 depends on a paired test, but the methods describe an unpaired design. Please clarify the analysis and report the test used. The revision locates the problem, states the evidence, and gives the authors a clear path to respond.

A second example concerns generalization. Instead of writing, The results are not generalizable, the trainee might write: The sample is limited to one institution, so the scope of inference is unclear. Please state the limits of generalization or add a limitation paragraph. The mentor’s annotation can explain that the original comment is a verdict, while the revision is a request tied to the manuscript’s evidence.

A third example concerns citation. Instead of writing, The introduction ignores prior work, the trainee can write: The introduction does not cite studies of the relevant topic; please situate the claim relative to that literature. The mentor can point out that the revision avoids implying negligence and tells the authors what kind of engagement is expected. These examples are not scripts. They are models of how to move from evaluation to explanation.

Professional Development, Not Free Labor

Co-review should be planned, credited, and bounded. It is not a way to offload a mentor’s review queue onto a trainee. Before starting, the mentor should confirm that the journal permits co-review and disclose the arrangement according to its policy. The trainee should know the manuscript’s confidentiality requirements and should not retain or reuse the text outside the review. The mentor remains responsible for the final submission.

At the lab level, co-review can be written into a professional development plan alongside conference presentations, paper drafts, and grant-writing experience. A realistic goal is one co-review per year for a trainee who is preparing for an independent research career. The annotated draft can be kept in a mentoring file as evidence of progress, provided it is stored securely and stripped of confidential manuscript content. When the trainee later lists reviewing experience on a CV, the mentor can confirm the scope of their contribution.

Guardrails for a Teachable Practice

Three guardrails make co-review defensible. First, confidentiality: no manuscript files or excerpts leave the approved review system, and annotations are stored only where the lab’s data policy permits. Second, disclosure: the journal’s policy on co-review and the trainee’s role should be followed exactly. Third, evaluation: the mentor should assess the trainee’s review on reasoning and tone, not on agreement with the mentor’s conclusions.

Peer review will remain imperfect, and no protocol will make it uniform. But a lab that annotates its reviews turns a private service obligation into visible instruction. The trainee learns how to read evidence, rank concerns, and write comments that are firm without being harsh. The mentor learns where their own standards are implicit. And the manuscript, ideally, receives a review that is more careful because someone had to explain why it was written that way.