Polishing a translation is mostly about not touching things. The instinct, when we open a draft with reviewer feedback attached, is to scan the whole document and fix everything that looks improvable. That instinct produces worse translations, not better ones.
Our job is targeted. The reviewer scored the draft and pointed at specific sentences, terms, or omissions. We address those, then we stop. The sentences the reviewer did not flag are not invitations to improve. They are decisions that have already been made and reviewed.
This is harder than it sounds.
The temptation of the unflagged sentence
Reviewers do not catch everything. They cannot. A reviewer reads a draft once or twice and makes a finite set of notes. Plenty of sentences they accepted could be sharper, more idiomatic, or better-cadenced. As we work through their feedback, we see those sentences too.
The temptation is to fix them anyway. We are already in the document. We can already see the issue. A small edit takes thirty seconds. What is the harm?
The harm is risk. Every change we make to an unflagged sentence introduces something that nobody has reviewed. Maybe our “improvement” subtly shifts the meaning. Maybe it breaks a parallelism that connected three paragraphs. Maybe the original translator chose that exact phrasing because the alternative collided with a term used later in the article. We do not know what we are about to disturb, because we did not write the draft and we did not review it.
When the next round of review happens, those untracked changes will either be caught and reverted, wasting the reviewer’s attention, or they will slip through unexamined. Neither outcome is good.
What the reviewer’s silence means
We have learned to read a reviewer’s silence the same way we read their notes. Silence on a sentence is a signal. It says: this passage met the standard, at least well enough that nothing else in the article was more urgent to flag.
That is not nothing. Reviewer attention is finite, and they spend it on the highest-impact problems. A sentence that survived a careful pass is presumptively fine. Treating it as fine is how we honor the reviewer’s judgment.
The same goes for the original translator’s choices. They selected a specific Ukrainian word for a specific English term. They chose where to break long sentences. They made hundreds of small decisions, most of them invisible. When we polish, we are editing a colleague’s work, not redoing it. The default move is to keep their decisions in place.
When we do touch the surrounding text
There are real exceptions. We change unflagged text when a flagged fix requires it for consistency. If the reviewer asks us to harmonize a theological term, we update every occurrence, not just the flagged one. If a flagged sentence was carrying important information that we have to redistribute, we adjust the neighboring sentence so the rhythm still works.
We also touch unflagged text when a flagged fix breaks something else. Fixing one sentence sometimes leaves the next sentence dangling or repetitive. In that case, the change is still scoped to the immediate neighborhood of the flag. We are not roaming the document.
The test we apply is simple. Does this edit follow from a reviewer’s note, or are we acting on our own judgment about a sentence the reviewer accepted? If it is the second, we leave it alone.
The discipline of stopping
We finish a polish faster than we expect, almost every time. Once the flagged items are addressed and the local rhythm is repaired, the work is done. The document is sitting open on our screen. The original English source is also open. Everything we touched, we touched on purpose. The rest of the article is the same article the reviewer signed off on.
This is the moment where over-polishing happens. A second voice says, “Let me just read through one more time and clean up anything else.” That second pass almost always produces edits we should not be making. Not because they are wrong in isolation, but because we are no longer operating within the brief.
So we close the document. We update the output, we write a short note describing what we changed, and we hand it back. The translation that comes out of a disciplined polish looks, on the surface, almost identical to the one that went in. That is the point. The reader will not see most of our work, because most of our work was choosing not to do work.
A lot of editing is like this. The visible labor is the changes; the invisible labor is the restraint. Engineers who review pull requests know the same temptation: while you are in the file, you can see ten other things you would do differently. Adding those to the PR muddies it, slows the review, and risks regressions in code nobody asked you to touch. A focused polish, like a focused PR, is easier to review, easier to roll back, and easier to learn from. It also keeps trust intact between the people producing drafts and the people refining them. When the original translator opens the polished version and sees that their voice is still recognizable, they know we read them with respect, not with a red pen.