Yet another conflux of words, this time “parent/child, grandparent/grandchild, boy/girl and baby”
On the Quenya side, we have Q. nostari “parents” from LotR, so we really have no choice but to use nostar for “parent”. We also have ᴹQ. “male and female parents” ontare/ontaro from the Etymologies, which may also remain legitimate, as well as N. odhron/odhril. There doesn’t seem to be a gender-neutral word for “parent” in Sindarin/Noldorin, but maybe odhron can be used for that purpose. All of these words are from derived the invertible root NŌ/ONO “to beget”.
For child we have Q. hína and S. hên, both from KHIN. There aren’t any real alternatives on the Sindarin side, but for Quenya we also have Q. onna from NŌ/ONO and ᴹQ. selda from SEL. I personally would avoid both of these: Q. onna seems to mean “begotten thing”, which doesn’t feel right to me, whereas selda “child” conflicts with selde “daughter” (but see “boy” below). If I used it at all, I would use Q. onna more with the sense “offspring”.
For grandparents we only have early words. On the Quenya side we have ᴱQ. haru/haruni for “grandfather, grandmother” from the 1910s. Their etymology is unclear, but phonetically they are compatible with later Quenya and they don’t conflict with other words, so using them should be fine. We also have G. dâd/mam, for “grandfather, grandmother”. These could plausibly be derived from the roots AT(AR) and AM for “father/mother”, perhaps via reduplication.
For “grandchild”, in Quenya we have only ᴹQ. indyo “grandchild, descendent” from the Etymologies. The only alternative is the very early use of yondo as “grandson” instead of “son”, which would be too confusing. The cognate for ᴹQ. indyo appears only as ON. ango, which would become (unattested) N./S. *ang. Since this is also the Sindarin/Noldorin word for “iron”, it is clearly unsuitable, which is probably why Tolkien didn’t use it. That leaves us only with very early Gnomish sion/siel “grandson/granddaughter”. These might be connected to the patronymic suffixes -ion and -iel, though the significance of the initial s is unclear.
There are a couple choices for “girl”. There are late words Q. nette and S. neth from the same root NETH as the words for “sister”. We also have Q. wende and S. gwend “maiden” from the root WEN(ED). I’d use both, with nette/neth for young girls and wende/gwend for older but unmarried girls. It seems that seldē may have archaically meant “young girl”, but I would use it’s derivates selde/sell exclusively for “daughter”. There is also N. dess for “young woman”, but Tolkien said it fell out of use and marked it as archaic.
Finding a word for “boy” is surprisingly hard. On the Sindarin/Gnomish/Noldorin side we have only G. nogin “boy, lad, urchin” from the 1910s with no clear etymology, as well as deleted gontha “boy” that may be a cognate of ᴱQ. yondo (in Gnomish, initial y usually became g). Both are hard to fit into a coherent Sindarin paradigm.
On the Quenya side, we have ᴹQ. seldo “(male) child”, but that is problematic for the same reason as selda above. All of primitive seldō, seldā, seldē would produce sell in Noldorin/Sindarin. In a couple places in his later writing, though, Tolkien indicated primitive yondō and Q. yondo may mean “boy” as well as “son” (VT47/26-27; PE17/190). Although S. ion was never glossed this way, it might be possible to use it for “boy” as well.
One possible scenario is that the root SEL may have originally meant “child”, with male, female and neuter variants, but the female variant early on became used as “daughter”. In Quenya, the neuter variant selda fell out of use and was replaced by Q. hína “child” but perhaps seldo “male child” = “boy” survived, while in Sindarin the only survival was S. sell “daughter”. Similarly, if primitive yondō originally meant both “son” and “boy”, and perhaps it retained both meanings in S. ion, but in Quenya came to mean primarily “son” with seldo being used mostly for “boy”.
Finally, for “baby”, on the Sindarin side we have only S. gwinig, a diminutive form derived from WIN “young”. We have several Quenya words for “baby” derived from the same root, wine, win(i)ce, winimo, all marked archaic. Of these I like winimo the best, but in modern Quenya it would likely be pronounced vinimo (since w usually became v). Another choice might be hinye “baby”, which seems to be a diminutive derived from KHIN “child”, but that seems a bit obscure. A lot of writers also use ᴹQ. lapse “babe” and ᴺS. laes (adapted from N lhaes) with the same meaning; I think these are fine as synonyms.
The net result is:
“parent”: Q. nostar (or alternately ᴹQ. ontaro/ontare), S. [N.] odhron (feminine odhril)
“child”: Q. hína, S. hên; or Q. onna “*offspring”
“grandfather/grandmother”: Q. [ᴱQ.] haru/haruni, S. [G.] dâd/mam
“grandchild”: Q. [ᴹQ.] indyo (feminine indye?), S. [G.] sion/siel (male and female)
“girl”: Q. nette, S. neth; or Q. wende, S. gwend “maiden”
“boy”: Q. seldo (and perhaps yondo), S. ion (the same word as “son”)
“baby”: Q. vinimo, S. gwinig; or Q. [ᴹQ.] lapse, S. laes “babe” (N. lhaes)
NOTE: In a previous version of this post, I wasn’t quite ready to accept Q. vinimo and lapse, or ᴺS. laes, but I’ve changed my mind based on how often people seem to use them.
Although widely accepted, ᴺS. laes is somewhat problematic phonetically: under normal Sindarin phonetic development, it might more properly be laus. But that’s a more complicated topic, one I’m not ready to delve into yet.