My, oh, my! Ask to learn: I never suspected that there would be so many codes for dead languages!
Now, a short and perhaps silly question: would it work just to mark those words (or roots, mostly) as <span xml:lang="pied">*k’ṃtom</span>, i.e., with a non-existing language code, or it would it sound all the bells in epubChecker? Really, the author is hardly ever giving words in a particular language, but rather pointing out 'cognates' or common roots in related languages... It seems rather an overkill to insert more than thirty different <span xml:lang="---">...</span> for mere fragments, mostly not belonging to actual languages, when what is really required is the notice 'don't treat this as a common word'.
Anyway, my thanks for the link to the code tables: good to know that there is such a complete reference for out-of-the-way languages.