The proper response to a word limit from an academic publisher is: Go
fu fly a kite in a lightning storm. However, since publishers are still clinging to the printed paper model and untenured folks are reluctant to play brinksmanship with editors who would cave in, we can find ourselves in the intellectually bankrupt circumstance of producing a word count for our manuscript. Those of us who use a type setting environment, like LaTeX, are in a somewhat challenging circumstance, and generally root around online looking for posts like this one that recommend how such a silly metric might be produced.
At this writing there are several options. You can install a user contributed package called wordcount. I have not tried that. You might use pandoc to convert the .tex file to a .docx format and then use the word count feature in Open Office / Libre Office / MS Office. You can also download TeXcount. If you are a linux user you can do the following:
“sudo apt-get install untex“ then “untex target.tex > count“, then “wc -w count“.
If your experience replicates mine you will get multiple values (often differing by a rather large amount). What do do? You could report the distribution, report the mean, or choose the low value. I have tried each (in different circumstances). Why? When faced with a silly institution I like to respond with silly behavior. But that’s me.