Sombody did something to remove your Zotero field codes. If you have a recent incremental backup of your document (that contains the field codes), I recommend using the backup, Copying your new text to the backup and re-adding the references that aren't live.

There are several ways that field codes can be removed from a document.

You can issue a command to remove them.

You can save your document to a new file format.

You can send your document to someone who opens your document in a different word processor and they save the document in a different format before returning it to you.

I've probably overlooked other ways this can happen.

I have several recommendations and comments. It is allways good practice to save incremental backups of your working document. I try to remember to save-as my document at least every 2 pages of work. MyDocument02, MyDocument03, MyDocument04, etc.

I never, ever share the most recent version of a document without saving it as a new version that incorporates an indicator of who received it -- usually I include the recipients' initials. For example, MyDocument04-abc. This way, even if the file when it is returned to me has not lost its field codes, I can easily distinguish my working document from the copy I received with comments. Another option is to send a pdf document not a word processor version.

Even if the worst happens and you do not have a backup with Zotero field codes, adding codes is not going to require many hours of effort. If your document uses a numbered style, use a print-out to replace the reference number with the proper Zotero reference. If you use an author (date) style you will need to spend more time getting the Zotero reference settings to match the sentence structure. I learned from my own mistake when I had to reinsert 400+ citations at the final stage of writing my doctoral thesis. Once I started, the process didn,t take too much time. The time wasn't altogether lost. The careful rereading process allowed be to find and correct typos and to clarify a few ambiguties I hadn't noticed with the normal reading process.

Good luck.

Any reason you're saving as .doc? And how are you establishing that the references are unliked?

Yeah, that definitely sounds unlinked. I'd use .docx, which is a much more stable and reliable format. Obviously this should work in .doc, but with almost everyone using .docx these days, it gets much more robust testing.

(Also, as you've noticed yourself, because it's an open XML format and not a binary, it's much easier to troubleshoot).