I'll look at my data, think about it hard, and come up with a solution.
Probably each cognate entry will start with the language name abbreviation (Lat, TochA, Grk, OAv, et al.) followed by the citation form, followed by the gloss; if there are two forms for a single language, I can follow the first entry with a comma and append another form and gloss. Each language entry is terminated with a line break or semicolon.
The real trick will be putting the footnotes in the right place.