What changes should be notated in a journal resubmission?

Question

I am currently in the process of submitting a manuscript to a journal. During this process my manuscript has come back for revision, and the journal has asked for an updated manuscript with revisions annotated. What level of changes should be annotated? Similarly is it acceptable to provide a generic comment when changes are for flow and not for content (e.g. "sentence rephrased for clarity")?

As an example (in my perceived order of required-ness):

• Adding or removing significant content (figures, paragraphs, sentences)
• Rephrasing a sentence changing the meaning
• Rephrasing a sentence changing the grammar, but not the meaning
• Adding/removing words that are superfluous (or change the meaning in an insignificant way)
• alteration of comma's, periods, etc.

Answer 1

An editor once "requested" that I submit a "tracked changes" or latexdiff version of my resubmission if I wanted him to deal with it in a timely fashion. Since then, my strategy has been to submit:

• revised manuscript (no annotation);

• latexdiff of revisions against version submitted last time, to highlight every last comma changed;

• cover letter, describing all scientific changes plus any major textual changes ("we substantially rewrote the methods section in an attempt to improve its clarity"). I refer the editor to the diff for minor textual corrections.

No editor has complained yet... ;)

Answer 2

What editors (and their journals) expect may vary. In the journal I edit, and also journals with which I am familiar as author and reviewer, expectations cover the two first points in your list. It is very common to request a point by point account for how reviewers' (and editor's) comments have been dealt with. This, to me is the important part of the revisions. Some editors want files highlighting changes, while other definitely do not. The important changes deal with the science and not the grammar or spelling. If the language has been a focal point for revisions, an editor will not likely check all changes you have made but rather read the manuscript to see if the language has been sufficiently improved.

While some (unclear what percentage) will not care about tracked changes, it is not wrong to supply them. Doing so allows the editor to chose which version (showing revisions or not) to use. But, always provide a clean version of the revised manuscript. I sometimes receive manuscripts with all changes visible in the manuscript and I feel uncomfortable accepting all on behalf of the author. The revised version is the responsibility of the author so the sharp new version should be included in the submission.