Proofreading of Latex document

Save your *.tex file as *.txt, say foo.tex as foo.txt. Tell the proofreader to open it in Word, correct whatever and -- this is the important step -- save it again as *.txt, let's say as foo-corrected.txt.

Just rename it into foo-corrected.tex and open it in your editor.

Obviously the proofreader will see all the commands. But just tell him to ignore anything that starts with a backslash and any line that starts with a percentage sign. For cases of doubt, send him the PDF: Everthing that is not visible in the PDF can be ignored safely.

I have had good results using Adobe Acrobat's File→Save As Other...→Microsoft Word→Word Document to convert a PDF produced by Latex. This gives reviewers a Word (.docx) document with remarkably similar appearance to the PDF. There are minor inconsistencies (spacing, hyphenation, and font selection problems) and some formulas don't come out quite right. But for content proofing without too much focus on layout it is perfect.

I have used this approach for a 200+ page report that has acronyms (glossaries package), chemical formulae (mchem), units and numbers formatted by siunitx, well-formatted tables (booktabs), cross references (cleveref), citations, a bibliography, and lots of figures. All of these elements look good in the Word document and even the hyperlinks from hyperref are preserved.

Acrobat has two layout options in the Save As Settings...: "Retain Flowing Text" or "Retain Page Layout". Both options work well. The latter produces a more exact replica of the PDF by using frames for the body text and other document elements. The former is probably the best option for reviewers as large changes to the text will cause problems with layout in the framed version.

I am using Adobe Acrobat XI Pro. Newer or older versions might produce different results. There may be free software or services that can do the same thing.

EDIT: Recent Experience with this Approach

Unfortunately, this approach produces formatting issues in the Word document that my coauthors find unacceptable. One issue is with tables, which are often malformed in Word. I could find no way to resolve this. I resolve a range of other issues by making changes to the Latex preamble before generating the PDF file for conversion:

• The PDF export strips out ligatures (e.g., for letter combinations ff and fi) and replaces them with a space (e.g., office becomes o ice). I turn off ligatures using the microtype package option \DisableLigatures. It may be better to not use microtype at all -- I have noticed some odd kerning of letters in the Word document that may be caused by microtype.
• I use small caps for acronyms and initialisms using the glossaries package. Small caps covert as lowercase in the Word document (Word does not use true small caps). I change the glossaries option \acronymfont to use upper case instead. Any other small caps (e.g., \textsc) will also convert to lower case.
• Hyphenated words in the PDF convert as hard-hyphens in the Word document so hyphenations can appear in the middle of the text block -- not on the margin. I turn off hyphenation using the hyphenat package.
• I have margin notes in my document but these convert to a second column in Word, which works poorly. I turn off margin notes by redefining the \marginline command.

This is my preamble code, which I uncomment before generating a PDF for Word conversion:

% Disable certain ligatures as many don't export to Word.
\DisableLigatures[Q,q,f,l,t,i]{}
% Make acronyms upper case as small caps don't export to Word.
\renewcommand{\acronymfont}[1]{\small\MakeUppercase{#1}}
% Turn off hyphenation as these will be "hard" coded in Word export.
\usepackage[none]{hyphenat}
% Turn off margin comments. They can produce strange results in Word.
\renewcommand\marginline[2][]{}

The results are imperfect, particularly for tables. I am reconsidering this work flow. It may be easier to originate the document in Word, collaborate in Word, and then produce a final document in Latex as the last step.

If your editor or external proof reader accepts standard pages as pdf you can use the stdpage package which creates fixed lines and characters per line. Just add \usepackage[linenumbers, hyphen=false]{stdpage} in your preamble.