How Editing Works
Editing is not proofreading.
Your dissertation passes through five sieves, from coarse to fine. Each one catches what the one above it let through, and proofreading is the last of them, not the whole process. Here is what each sieve catches, why you can’t see most of it yourself, and what I will ask of you along the way.
What comes before proofreading
Many students come to an editor expecting a proofread: a last read for typos before the file goes in. That is a real service, and it matters. It also comes last, after everything else is settled, because there is no sense polishing a sentence in a chapter that is about to change.
Before that, a dissertation needs a longer and more layered read. Is the structure right? Does every citation have a reference, and every reference a citation? Does a term mean the same thing in Chapter 2 and in Chapter 5? Does the document follow your university’s template? No single pass can answer all of those, so I don’t edit in one pass. I edit in five, each with a finer mesh than the last.
Proofreading is the last sieve, not the whole process.
| Proofreading | Editing (all five sieves) | |
|---|---|---|
| When it happens | Last, once the text is settled | From the first day, in a set order |
| What it looks for | Typos, spacing, stray punctuation, page numbers that moved | Structure, then citations and references, then language, then your university’s rules, and a proofread to finish |
| What it asks of you | Usually nothing | A few answers along the way, because some decisions are yours |
| What you get | A clean last page | A document that holds together from the title page to the final reference |
If you are a good writer, you may be thinking a proofread is all you need. Good writers need the earlier sieves too. What those sieves catch has little to do with how well you write, and a lot to do with what no writer can see in their own pages. More on that below.
Try it: is each task proofreading, or editing?
Editing or proofreading?
Twelve tasks from an imaginary dissertation. Sort each one into a pile.
The five sieves
Picture five sieves stacked in a funnel. The top sieve has the coarsest mesh and catches the biggest things. Each one below it is finer, and the last is so fine it catches only what the others left behind. Your file goes in at the top and comes out at the bottom clean. If you have not defended yet, there is also a sixth pass for after your defense; more on that below.
The order is deliberate. Each sieve assumes the one above it is finished, so nothing is polished twice and no later fix breaks an earlier one.
A paragraph to watch
To see the sieves work, here is a paragraph from an imaginary dissertation, a study of seasonal lifeguards at municipal pools. Everything in it is invented. It reads well, and it has problems no one would find in a single pass.
Click a sieve to see what it catches.
Click a sieve, or use Next, to watch one paragraph go through all five.
Prefer to read it? The five sieves, step by step
As written
Staff Retention
Lifeguards who completed the training clearly and strongly demonstrate higher retention than those who did not. In the 2021 survey, 62% of lifeguards said they returned for a second summer after the training began in 2022 (Table 4). The program led to a stronger sense of belonging among the guards, and staff described feeling valued. These findings echo Hallberg (2020), who studied a similar program.
Sieve 1 · Structure
On an estimate, this is called Phase 1, Mechanical & Structural.
What it catches. The coarsest mesh, and the biggest things: your university’s template, margins, real heading styles, the table of contents, front matter, and how tables and figures are built. Everything below this sieve stands on it, which is why it comes first.
In our paragraph. “Staff Retention” was typed in bold, so Word doesn’t know it is a heading. It would be missing from the table of contents and the navigation pane. It becomes a real Level 2 heading, and nothing else about the text changes.
Sieve 2 · Citations and conformity
On an estimate, this is called Phase 2, Conformity Checklists.
What it catches. Every in-text citation matched to a reference entry, and every entry to a citation. The reference list itself. Numbers, abbreviations, lists, quotations, tables and figures, checked against APA 7 and against each other.
In our paragraph. The text cites Hallberg (2020), but no Hallberg source is in the reference list, and I can’t find a 2020 study by that name. The text says 62%, and Table 4 says 61%. Only your copy of the source and your own data can settle these. Two queries come back to you, each asking you to look something up.
Sieve 3 · Language
On an estimate, this is called Phase 3, Line Edit.
What it catches. Grammar, spelling, punctuation, sentence and paragraph structure, and academic tone, with your voice intact, plus a sweep for patterns that repeat, such as a word used two ways.
In our paragraph. Three things would stop a careful reader. “Clearly and strongly demonstrate” and “led to” say more than a survey can show. The same people are “lifeguards,” “guards,” and “staff.” And a 2021 survey can’t report on something that began in 2022. Each comes back to you as a query, and your answer decides.
Sieve 4 · University rules
On an estimate, this is called Phase 4, University Compliance.
What it catches. Your program’s own checklists, template rules and formatting conventions, line by line: what your university’s final style review will check.
In our paragraph. The last sentence compares your results with Hallberg’s. Your program’s template keeps the findings chapter to your own findings, so that comparison belongs in Chapter 5. Moving it is a choice for your committee, so it comes back as a query for your chair.
Sieve 5 · Proofreading
On an estimate, this is called Phase 5, Final Check & Handoff.
What it catches. The finest mesh, and the last. A final consistency check over the whole document, then a proofread of everything that moved during the earlier passes.
In our paragraph. By now your answers are in and the text has shifted. This sieve catches what the changes left behind: the new heading’s page number in the table of contents, a doubled space where a sentence moved out, and one last check that every year and table number still agree.
The same paragraph, after five sieves
After five sieves, with your answers in
Staff Retention
Lifeguards who completed the training were more likely to return than those who did not. In the 2022 survey, 61% of lifeguards said they returned for a second summer after the training began in 2021 (Table 4). Training was associated with a stronger sense of belonging, and lifeguards described feeling valued.
Your answers finished it: the source details from your copy of the report, the right figure from Table 4, the years of the survey and the training, your chair’s call on Chapter 5, and your yes to two suggested phrasings. The Hallberg comparison now lives in Chapter 5, with its year corrected and the report added to the reference list. Every decision about what the study found stayed with you.
In a real file, some of these problems would be caught by more than one sieve. That overlap is the point.
Move the slider to see roughly how much there is to check in a file your size.
What gets checked in a file your size
Move the slider to your word count, from Chapter 1 through the reference list.
Rough counts, based on the averages across full dissertations I have edited. Every file is different: the ranges show the lowest and highest I have seen, scaled to your word count.
After your defense: a sixth pass
Passing your defense is usually not the last edit. Your committee approves the work with required revisions, and you make them yourself, often quickly and often tired, at the end of a very long process. Those revisions are new writing dropped into a formatted document, and they reliably bring back the kind of problems the five sieves caught: a new source that never reaches the reference list, a heading typed instead of styled, a table of contents whose page numbers have shifted, a new table that renumbers every table after it.
Your university then runs its own final style check, and at many institutions that approval is what releases your signature page and lets you graduate on time. That is why there is a sixth pass, the post-defense check. It is optional, and it sits outside the funnel because it happens after your defense: a short look at what changed, so the file is in line with your committee’s revisions and your university’s format before it goes to the final check. If you would rather do that last look yourself, that is fine too. What matters is that somebody looks, with formatting eyes, after the last content changes.
If your defense is still ahead, Defense Prep covers the weeks leading up to it and mentions this step too.
How professional publishing works
This is not a method I invented for dissertations. It is how published work is made.
A journal article or a scholarly book does not go from the author’s desk to the reader in one step. Peer reviewers question the argument. The author revises, often more than once. A copyeditor works through the language, the references and the consistency of the whole. A proofreader reads last, after the text is set, for the small things that crept in along the way. Each stage has a different reader with fresh eyes, because the author is the one person who can’t read the work as a stranger would.
A dissertation asks one person to do all of that alone, in a single draft, at the end of a very long project. Your committee plays the part of peer review: it questions the argument, and those decisions are yours and theirs. I come in on the form, as copyeditor first and proofreader last. I edit form, not findings.
My own book, Synchronicity in Grief, went through the production stages with Routledge. I submitted the manuscript in June 2026. The copyedit came back with thirty-one author queries across a hundred thousand words, and I was the one being edited. I returned the copyedited files four days before my deadline. Then came page proofs, which I could review only once, and an index I built by hand over thirty hours. In those proofs I found a note to myself from an early draft. It had survived my writing, my submission and the copyedit, and I could not see it until the pages looked like a book. The book came out in October 2026.
I wrote about each stage as it happened: how the book began, the copyedit queries, page proofs and the index, and the last thing I did to the book.
Why you can’t see your own errors
You wrote it, so you know what it means. That is exactly the problem. When you reread your own sentence, your mind supplies what you meant to say, and the page you see is the page you intended. A missing word, a term used two ways, a citation whose year doesn’t match its reference: these pass right under your eye.
It isn’t carelessness, and it isn’t tiredness. It happens to every writer, editors included. It is why published work has several readers.
There is a second reason, which is scale. A dissertation is too big to hold in your head: five chapters, a long reference list, tables, figures, and a template with its own rules. No one can track a single term across a hundred pages, or check every citation against every reference from memory. That is the work I do. I read for these things, with checklists built for them, from the first page to the last.
And your committee is reading for your argument. Detail is a different kind of reading, and they shouldn’t have to do it.
So when your file comes back marked up, the marks are not a verdict on your writing. They are what a second reader sees that the author can’t.
Each sentence has one problem. Can you find it? It is harder than it looks.
Find what I found
Eight sentences, one problem in each. Click the part you think I would query.
Five kinds of query, and your side of the work
As I work, I leave questions for you in the margin of your file, as comments. I call them queries. I leave one wherever a fix would change your meaning, because I would rather ask than guess.
You will see a lot of them, and that is normal. Comments are a measure of how closely I read, not of how well you write.
Every query asks one of five things of you.
Check your source and send me the detail.
QUERY: Hallberg (2020) is cited here but isn’t in the reference list, and I can’t find a 2020 study by Hallberg on this topic. Which source did you mean? Would you send me the details from your copy?
Tell me which you meant.
QUERY: This paragraph calls the same group “lifeguards,” “guards,” and “staff.” Which term do you want throughout? If it is “lifeguards,” I’m happy to carry it through the chapter.
Take it to your chair or mentor, then tell me what they say.
QUERY: Chapter 1 lists two research questions and Chapter 3 lists three. That choice belongs to your committee, not to me. Would you ask your chair which set is current? Once you know, reply here and I will line up the wording.
Yes or no is enough.
QUERY: A survey can show that training and retention go together, not that one caused the other. Would “was associated with” work in place of “led to”? If that is faithful to your data, I’m happy to drop it in.
Do something I can’t do in the file.
QUERY: Figure 2 is a picture, so I can’t move the label that sits on top of the bars. Would you re-export it from your software with the label moved? Once the new image is in, I will check the caption.
When a query says “your call,” either answer is fine. I am offering an option, not correcting a mistake, and “keep it as is” is a perfectly okay response, too.
Your side of the work
Editing is a partnership, and part of the work is yours. Queries come in every phase, not only at the end, so plan some time for them in each one. Most need a look-up or a decision only you can make. Four habits make it go faster:
- Keep your source PDFs and notes where you can reach them, so a look-up takes minutes instead of an evening.
- Tell your chair now that you may bring them a few questions during the edit, so their answers don’t hold up the schedule.
- Reply inside the comment. One line is enough: “Yes, change it.” “Page 112.” “My chair says keep it.” If you need time on one, write “checking.” If you want different wording, type it in your reply and I will place it.
- Leave the tracked changes alone, and work in the one file I send you, not a copy. I accept the changes at the right stage, so nothing is lost and the file stays one clean master.
Try it: four queries, three possible replies each. Which reply helps me most?
Practice answering a query
Four queries, three possible replies each. Which reply helps me most?
What happens if a query isn’t answered
Sometimes a query waits: the source is in a library you can’t reach, your chair is traveling, or the week got away from you. That is all right, and it does not stop the edit. I keep working on everything that doesn’t depend on your answer.
What I do not do is guess. If a fix would change your meaning, the text stays exactly as you wrote it until you tell me what you want.
And nothing gets lost. Every query still open carries forward and appears in the feedback letter that comes with your final file, so you can see what is still waiting on you and what each one needs.
One critical thing: a query left open is a spot left as it is. A missing page number is still missing when the file reaches your committee or your university’s style review, so the sooner an answer comes, the cleaner the finish. If you need time on one, write “checking” in the comment. That helps me know you are looking into it and keeps the project on track.
Not sure what your file needs?
Bring your university, your timeline and your questions. In a strategy session, we talk through where your dissertation is now, which of the five sieves it needs, and what the next step is.