This root and ones likes it were the basis for the initial element of the name S. Felagund for much of Tolkien’s life. The name first appeared (untranslated) in the Lays of Beleriand of the 1920s (LB/80), but did not have a derivation until The Etymologies of the 1930s. In that document, it was based on ᴹ√PHELEG “cave” which had derivatives like ᴹQ. felya and N. fela of the same meaning (Ety/PHÉLEG).

In Notes on Names (NN) from 1957, Tolkien first gave ✶phelgā > Q. felya/S. fela “mine, boring, tunnel, underground dwel[ling]” matching the entries from the earlier The Etymologies, but then revised these forms to Q. felco and S. feleg “cave, mine, underground dwelling” based on new forms of the root: √PHELEK and √PHELES (PE17/118). In The Shibboleth of Fëanor from 1968 Tolkien explained S. Felagund as an adaptation of the Khuzdul title Kh. Felakgundu “Cave-hewer” (PM/352). However, in notes from 1969 Tolkien gave a primitive form philig- and as the basis for S. fela, derived from primitive ✶philga or ✶phelga, but later in the same note he gave the ancient plural form as ✶phelgai so he seems to have come full circle back to *√PHELEG (NM/304).

In this 1969 note he described the meaning of a phelga as follows:

It was used of minor excavations made by wild animals as dens or lairs, and also as temporary dwellings by wandering folk, Dwarvish or Elvish; it was usually distinguished from the larger caves of geological formation used and extended by stone-workers. It was thus naturally used of the “setts” of badgers (which seem to have existed in great numbers in parts of Beleriand).

Neo-Eldarin: For purposes of Neo-Eldarin, I’d use √PHELEG as the form of the root and thus use the 1930s forms of its derivatives which were restored in 1969, though it does seem that it may have been Sindarin-only in its later iteration. I’d allow Q. felya/S. fela to mean any of “den”, “underground dwelling” or “mine”, but limit it to intentional diggings rather than natural caves. I would also retain S. feleg “cave” as a Sindarin loan word from Khuzdul (but not Q. felco, as it is too far removed).

See *√PHELEG for discussion.