Elsevier tells you to select the whole document with Ctrl+A and press Ctrl+Shift+F9 to strip your reference manager's hidden codes before submitting.
Microsoft documents that shortcut as: "Unlink a field."
Not "unlink a citation". A field. And Microsoft's own field-code reference lists TOC, SEQ and REF as fields — your table of contents, your automatic numbering, your cross-references.
So the published instruction, applied to a whole-document selection, has no citation-only boundary anywhere in it. Nothing in the documentation gives you a reason to expect it to leave those other objects alone.
I went looking at this because a lot of people run that command in a hurry, near a deadline, on the file they are about to submit — and because the thing everyone thinks they are solving by running it is not actually a problem.
Let me get that part out of the way first, since it is the usual reason people start down this road.
A reference list is machine-generated, identically formatted and structurally repetitive, so people assume it is dragging their AI percentage up. It is not, on either side. Turnitin's AI writing report analyses only what it calls qualifying text — prose in standard grammatical sentences — and a release note dated 9 August 2023 separately says bibliographies are excluded when the report is processed.
If a bibliography is coloured in a report you are holding, the first thing to check is whether you are looking at the Similarity Report rather than the AI writing report. Matching reference entries are what a shared citation style is designed to produce.
So the expensive step is not the bibliography. The expensive step is the irreversible command someone told you to run.
Here is what I found comparing the vendors' own pages. Three reference managers, four add-ins, four different commands — and they do not behave the same way.
Zotero documents Unlink Citations, and calls the removal irreversible.
EndNote documents Convert Citations and Bibliography → Convert to Plain Text, and explicitly says the command creates a new copy for you. Clarivate's wording on going back is worth quoting, because "irreversible" tends to get read as marketing caution:
"Field code removal is a one-way process. Once you've removed EndNote™ field codes, you will need to reinsert the citations or revert to a backup of the document that still has the field codes in it."
Mendeley ships two different add-ins, and this is where it gets interesting. The older Citation Plugin used fields and had an Export command. Mendeley Cite, the newer one, uses different object language entirely: it calls a citation or bibliography a content control, and tells you to right-click the object and choose Remove Content Control.
That difference matters, because Microsoft's field-shortcut page says nothing about what Ctrl+Shift+F9 does to a content control. If your citations are content controls, the shortcut everyone recommends is documented against a different object type than the one in your document.
Zotero adds one more axis: its Document Preferences let you store citations as Fields or Bookmarks, advising that unless you need to collaborate with LibreOffice users, "you should always choose Fields". So even within one manager, the object in your file is a setting.
Fields, bookmarks, or content controls. Three possibilities, three different removal paths, and one keyboard shortcut that people apply to all of them.
There is a free, instant, completely reversible way to find out what is actually in your file before you run anything irreversible. Same Microsoft page:
"Switch between all field codes and their results. Alt+F9"
Press it. Word toggles between field codes and their displayed results. A table of contents exposes a TOC instruction; numbering exposes SEQ; a cross-reference exposes REF. If your citations also show up as field codes, fields are involved. If they don't, that tells you something too — and you should go read your add-in's own page rather than assume.
It costs one keystroke and it is the only step here you can undo by pressing the same keystroke again.
One more thing, for anyone whose plan is to reword bibliography entries until they stop matching.
Set aside whether that is a good idea. Just look at what the software does with the edit. Zotero's page is blunt:
"Manual edits made to the bibliography in Word will be overwritten the next time Zotero refreshes the document."
And for in-text citations, on the same page: manual modifications prevent Zotero from automatically updating that citation.
So in a document with live fields there are exactly two outcomes, and neither is the one you wanted. Either the next refresh silently reverts your edit, or the edit sticks and that one citation is now disconnected from the library — meaning the next style change updates every entry except that one.
The order of operations that keeps this cheap:
- Keep a working master. Name the delivery copy separately before removing anything. EndNote makes a copy for you; the others do not say they do.
- Finish all citation work first — style changes, added sources, page numbers, prefixes. After flattening, none of it can be done by the manager any more.
- Refresh once before flattening, so what gets frozen is current output rather than something stale.
- Identify the object, then follow that product's own path. Zotero: Unlink Citations. EndNote: Convert to Plain Text. Mendeley Cite: Remove Content Control. Not one generic shortcut for all three.
-
Press
Alt+F9first, and again on anything that comes back to you from any other tool, to confirm your fields are still fields.
And the honest boundary on all of the above: I read the vendors' documentation, I did not run a Word-version test matrix. What I can say is what the pages say — and what they say is that the command is scoped to fields, that at least one major add-in does not use fields, and that going back is not offered.
Originally published at HumanPen: https://humanpen.net/blog/zotero-endnote-citations-and-ai-flags?utm_source=devto&utm_medium=article&utm_campaign=D214
Top comments (0)