A noun appearing as ᴹQ. anta “face” in The Etymologies written around 1937 under the root ᴹ√ANA “to, towards” (Ety/ANA¹), based on an extension of this root: ᴹ√ANAT (EtyAC/ANA¹).
Conceptual Development: The earliest appearance of this word was as ᴱQ. anta “jaw” in The Qenya Phonology of the 1910s, where it was derived from ᴱ✶mtā, related to ᴱ√MATA “eat” (PE12/26). It was mentioned again in the contemporaneous Gnomish Lexicon as a cognate to G. ant “cheek” (GL/19), but in the contemporaneous Qenya Lexicon and Poetic and Mythological Words of Eldarissa it was ᴱQ. anto “jaw” (QL/31; PME/31). ᴱQ. anta reappeared in Early Noldorin Word-lists of the 1920s, but there it had the gloss “cheek” and its Gnomish cognate G. ant was “face” (PE13/137, 160). In Early Qenya Word-lists from the 1920s, however, anta retained the gloss “jaw” (PE16/136).
As noted above, in The Etymologies ᴹQ. anta had the gloss “face” and a new derivation from ᴹ√ANA “to, towards” (Ety/ANA¹), perhaps meaning “✱front of the head”. In that document Tolkien introduced ᴹQ. anka for “jaw” based on ᴹ√NAK “bite” (Ety/NAK). In notes on The Feanorian Alphabet from around this time, he revised the gloss of ᴹQ. anta from “jaw” to “face” (PE22/21 note #64), which was followed by a chart that had both anta “face” and anka “jaw” (PE22/22). However, he then wrote “jaw” faintly above anta and marked through the gloss of anka (PE22/22 note #67).
In version of these notes on The Feanorian Alphabet from the 1940s, Tolkien had anta “jaws”, but there it was revised to ᴹQ. anto “mouth” (PE22/50 and note #185). In the version of Tengwesta Qenderinwa from around 1950 (TQ2) Tolkien again had Q. anta “jaw” < ✶amtā based on the root √MAT, but he revised the primitive form to ✶ankā and then marked through the entire paragraph (PE18/85 note #72). In the tengwar charts from Appendix E of The Lord of the Rings, Tolkien had Q. anca “jaw” and Q. anto “mouth” (LotR/1123).
To summarize, it seems Tolkien revised the meaning of word anta from “jaw” >> “cheek” >> “face” in the 1910s through 1930s; then in the 1940s and 50s he considered restoring anta “jaw” but ultimately settled on Q. anca “jaw” and Q. anto “mouth”.
Neo-Quenya: Given the ultimate result above, I think it is possible that the sense ᴹQ. anta “face” < √ANA “to(wards)” from The Etymologies remains viable, and I would use the word anta with that meaning for purposes of Neo-Quenya. I would also let it retain its ᴱQ. meaning “cheek”, as we have no other Quenya words with this sense.
anta- (1) vb. "give" (ANA1, MC:215, 221), pa.t. antanë (antanen "I gave", VT49:14) or †ánë, perfect ánië (PE17:147, cf. QL:31). According to VT49:14, Tolkien noted that anta- was sometimes often with an "ironic tone" to refer to missiles, so that antanen hatal sena "I gave him a spear (as a present)" was often used with the real sense of "I cast a spear at him". Usually the recipient of the thing given is mentioned in the dative or allative case (like sena in this example), but there is also a construction similar to English "present someone with something" in which the recipient is the object and the gift appears in the instrumental case: antanenyes parmanen, "I presented him with a book" (PE17:91). The verb occurs several times in FS: antalto"they gave"; strangely, no past tense marker seems to be present (see -lto for the ending); antar a pl. verb translated "they gave", though in LotR-style Quenya it would rather be the present tense "give" (pl.); antaróta "he gave it" (anta-ró-ta "gave-he-it"), another verb occurring in Fíriel's Song, once again with no past tense marker. Also antáva "will give", future tense of anta- "give"; read perhaps *antuva in LotR-style Quenya; similarly antaváro* "he will give" (LR:63) might later have appeared as antuvas (with the ending -s rather than "Qenya" -ro for "he"). Antalë imperative "give thou" (VT43:17), sc. anta "give" + the element le "thou", but this was a form Tolkien abandoned. Apparently ana** was at one point considered as another imperative "give", but Tolkien rewrote the text in question (VT44:13), and the normal patterns would suggest *á anta with an independent imperative particle.