Tolkien's shifting ideas about his languages are well illustrated by a
detail in the linguistic evolution from primitive Elvish to Sindarin. This is
not a question that has only academic interest: Some of the "Noldorin" words in
the Etymologies should be corrected before they are used by people
writing in mature Sindarin - the words must so to speak be updated to comform
with Tolkien's final decision in this matter.
The question before us (or rather before Tolkien) is this: Does primitive *ai become oe or ae in Sindarin?
Sounds simple, don't you think? Especially when you are inventing the language yourself and can just pick one or the other? Wrong! It wasn't simple! Tolkien needed decades to sort out this question! He must have spent many, many sleepless nights trying to decide whether he wanted the word for, say, "sad" to be noer or naer.
The Etymologies reflects Tolkien's indecision. In many cases, primitive *ai becomes ae in "Noldorin" (> Sindarin) words. For instance, the stem for "(small) bird", AIWÊ, yields the Noldorin word aew. For other examples of AI (AY) becoming ae in Noldorin, see the stems DAY, GÁYAS, KAY, KAYAN/KAYAR, LAIK, NÁYAK, TAY, WAIWA, YAY, to list some of the more obvious cases. But sometimes, primitive *ai becomes oe instead. For instance, the stem SPAY yields the word foeg "mean, poor, bad", not faeg, though the Quenya cognate is faika and the primitive form must have been *spaikâ. Sometimes, Tolkien's indecision is glaringly obvious. From the stem NAY, Tolkien derived the word noer "sad, lamentable", but two seconds later he had the same stem yielding nae "alas". Why not noe - or alternatively, why not naer? Under GAY, *gairâ (my reconstruction) yields both gaer and goer, as if Tolkien could not make up his mind and even considered the possibility of several dialects - one dialect in which older *ai became ae, and one in which it became oe. In the entries SLIW and MIL-IK Tolkien first had primitive words with *ai yield "Noldorin" words with ae, but on second thoughts he changed ae to oe (e.g. *slaiwâ first yielding thlaew "sickly, sick, ill", which he then changed to thloew).
The Etymologies was written in the thirties. Would you believe that a
full twenty years later, Tolkien was still struggling with the
ae/oe-question? When he was writing the Appendices to LotR, his indecision was very obvious. The name Dirhael was at one point changed to Dirhoel (PM:263), signaling the beginning of an oe-phase. The Noldorin/Sindarin word for "summer" (Quenya Lairë) was originally given as Loer in an early version of the Appendix on calendars (PM:135). But now the awesome decision had to be made: Once LotR was published, containing Sindarin words with either oe or ae, this feature of the phonology would become fixed and deprive Tolkien of the possibility to ever change his mind again. So now he had to make absolutely sure that he had picked the right choice. After all, he would have to live with it for the rest of his life!
There is probably no actual record of this hard day in his career as a language-maker. But let us imagine Tolkien taking a long walk trying to figure it out, carefully weighing the alternatives. Deadline was approaching; a decision had to be made. Ae or oe, oe or ae? Which diphthong suited the Elvish tongue better? What was an Elf more likely to say? Suddenly Tolkien decides to put the question out of his mind. -Come on, he tells himself, you can't go on tinkering with the Elvish tongues forever! You have used oe-forms in the Appendices you are preparing for The Lord of the Rings; just leave them alone! You don't really think the readers will care, do you?
Then Tolkien catches sight of a small bird sitting in a tree nearby. It looks back at him. Suddenly he gets a feeling that it is accusing him of something. Some hesitant steps bring Tolkien closer to the tree. The bird is still watching him, but it does not fly away. He feels an inexplicable urge to have a closer look at it. Now was that bird in Elvish terms an oew? No! No! Suddenly the light breaks through in a great flash! It had to be an aew! There was simply no way it could be an oew! How could he have been so wrong all these years?
The bird looks at him with approval, then flies away. Tolkien's indecision is gone. He feels calm and yet eager. As soon as he gets back home, he changes the word for "summer" from Loer to Laer (PM:136). And from now on, primitive *ai did yield Sindarin ae, never oe. (See for instance Letters:282, where Tolkien has *laikâ yielding Sindarin laeg "green"; this was written in 1958.) It took Tolkien more than two decades of inner struggle to figure out what the Elves really said, but at last he got it right!
However, it may be that one oe-form made it into LotR after all. The
name Nen Hithoel, the lake near the falls of Rauros, may have been
intended to mean "Mistpool-water" when Tolkien first made it (hith +
oel). The first two volumes of the LotR trilogy had already been
published when Tolkien at last reached his final decision in the
oe/ae-question, and Nen Hithoel is only mentioned in the first of
them. But now, the word for "pool" was ael, not "oel" as it had been in
the Etymologies. Perhaps Tolkien, noting to his dismay that one of the
now obsolete oe-forms had sneaked into his magnum opus, decided to
reinterpret the name. (Note that the diphthong oe as such was not
banished from the language; it still has its rightful place in many other
words, where it is not derived from earlier *ai.) In the Index to
Unfinished Tales, entry Emyn Muil, Nen Hithoel is
translated "Mist-cool Water". I suspect that this was not the meaning Tolkien
originally intended for this name; nowhere else in the published
material is any element even resembling oel assigned the meaning
Whatever the case, some names in the Silmarillion confirm Tolkien's decision that *ai should yield Sindarin ae, not oe. The Great Sea is now called Belegaer, not "Belegoer" as in the Etymologies (see BEL, AYAR/AIR, ÁLAT). The Sindarin name of the Misty Mountains, Hithaeglir, contains aeglir "range of mountain peaks" - which was "oeglir" in the Etymologies, stem AYAK. There is also the Aelin-Uial or "Meres of Twilight", which name was given as "Oelinuial" in the Etymologies, stem AY. Cf. also the Appendix, where Sindarin maeg "sharp, piercing" (as in Maeglin "Sharp Glance") is said to be the cognate of Quenya maika.
It should be noted, though, that while *ai now becomes ae in Sindarin, *ây with a long â still becomes oe! In PM:363, Tolkien derives Sindarin goe "terror, great fear" from primitive *gâyâ.
This, I hope, is a complete list of the "Noldorin" oe-words from the Etymologies that should have ae in mature Sindarin:
1. doer "bridegroom" (NDER; older form ndair given) should be corrected to daer.
2. thloew "sickly, sick, ill" (SLIW; primitive form *slaiwâ given); at first, Tolkien actually wrote thlaew, then corrected it. The correction should be ignored, since he later changed his mind again. The later form flaew was apparently never "corrected". However, neither thloew, thlaew or flaew should be used by people writing in LotR-style Sindarin. We must go for lhaew, since another revision separating "Noldorin" from Sindarin affected the development of primitive initial sl-: Tolkien later decided that this became lh- in Sindarin, not as in "Noldorin" thl-, later fl-.
3. foeg "mean, poor, bad" (SPAY; primitive *spaikâ, my reconstruction) should be faeg.
4. foen "radiant, white" (PHAY; primitive *phainâ, my reconstruction) should be faen.
5. goer "red, copper-coloured, ruddy" (GAY; primitive *gairâ, my reconstruction); Tolkien actually gave the form gaer as well, and this form is to be preferred. Do not confuse this with the noun gaer "dread" (GÁYAS)
6. lhoeb "fresh" (LÁYAK, primitive *laik-wâ) must become laeb. (Again, more than one revision is involved; while original initial *l- yielded lh- in "Noldorin", Tolkien later decided that this sound was unchanged l- in Sindarin.) But the words Lhoebelidh, Lhoebenidh "Greenelves" can be ignored altogether, since Tolkien later decided that Green-elves were called Laegil in Sindarin (sg. Laegel, class plural Laegrim, Laegel(d)rim). See WJ:385.
7. moe "soft" (older maiga) and the related word moeas "dough" (MASAG) are special cases: Emending moe to mae would produce a clash with the adverb mae "well" (as in Glorfindel's greeting to Aragorn, mae govannen "well met"). It may be practical to leave this word alone and simply assume that it has another etymology than Tolkien first intended.
8. moel "lust", whence the adjective moelui "lustful" (MIL-IK; noun from primitive *mailê, my reconstruction). Tolkien first wrote mael, maelui; then he entered one of his oe-phases and changed it: ignore the change. The related Noldorin name of Melko(r), Maeleg > Moeleg, can be ignored altogether: Tolkien later decided that the etymology of Melkor's name had nothing to do with the stem MIL-IK "lust, greed". In the Silmarillion Index, entry Melkor, the Sindarin name of the evil Vala is given as Belegûr, always altered to Belegurth "Great Death". Of course, he was normally called Morgoth.
9. noer "sad" (NAY; primitive *nairâ, my reconstruction). With the strengthened meaning "dreadful", the new form naer is apparently found in the compound naeramarth "dreadful doom" in the name Cabed Naeramarth in the Silmarillion. Cf. also naergon for "woeful lament" in PM:362. An alternative form of the name of Túrin's sister, Nuinoer, can be ignored since the Silmarillion has Nienor.
10. oeg "sharp, pointed, piercing" must be ignored: Tolkien decided that no cognate of Quenya aica "fell, terrible, dire" was used in Sindarin, "though aeg would have been its form if it had occurred" (PM:347). On the other hand, there is Aeglos "snow-point" as the name of Gil-galad's spear, so there must be a noun aeg "point". There is also the word aeglir (see oeglir below).
11. oegas "mountain peak" (AYAK), pl. oeges, is related to the previous word and must therefore become aegas pl. aeges instead.
12. oeglir "range of mountain peaks" (AYAK) is yet another related form; its new form aeglir is attested in the Silmarillion, in the name Hithaeglir (the Misty Mountains, translated "Line of Misty Peaks" in Christopher Tolkien's index).
13. oel "pool, lake", pl. oelin (AY, primitive *ai-lin-) is attested in its new (pl.) form aelin in the name Aelin-uial "Meres of Twilight" in the Silmarillion. The Silmarillion Appendix likewise gives aelin "lake, pool" (this is actually the pl. "lakes, pools": sg. ael).
14. oear "sea" (AYAR/AIR) - aear in the hymn to Elbereth in LotR. In the Silmarillion, the Great Sea is similarly called Belegaer, not Belegoer as in the Etymologies (BEL, AYAR/AIR, ÁLAT). In Letters:386, in a letter dating from 1967, the stem is said to be AYAR, just like in the Etymologies, and the Sindarin word is given as aear in a footnote. This, however, was not Tolkien's (apparently) final decision. In a text written at least one year later, published in PM:363, the Sindarin word for "sea" comes from a stem GAYA and is therefore gaear, gaer-, of which aear in the hymn to Elbereth would be the lenited form ('aear). So oear, oer in the Etymologies must be corrected to gaear, gaer, and for oeruil "sea-weed" (UY), we must read gaeruil. - Actually gaear instead of aear as the word for "sea" occurs already in the essay Quendi and Eldar from about 1960; see WJ:400. The 1967 letter quoted above indicates that Tolkien had then reverted to aear, but shortly after he once again decided to go for gaear. (Back and forth, back and forth...)
15. rhoeg "crooked, bent, wrong" (RÁYAK, primitive *raikâ, my reconstruction) must be corrected to raeg (with r for rh; this change reflects another revision of the "Noldorin"/Sindarin phonological development, forming a parallel to l for lh - see lhoeb, Sindarin laeb, above, or click here for a fuller treatise regarding this revision).