It is well known that au usually became o in polysyllables in both Sindarin and Noldorin, a process Tolkien mentioned several times:

... thus the ending au, ī, ū becoming thus unstressed was later reduced to o, i, u ... (Feanorian Alphabet, 1930s, PE22/31).

As the mutations following the preposition o show, it must prehistorically have ended in -t or -d. Possibly, therefore, it comes from *aud, with d of the same origin as that seen in Q. oär (see above). Some have thought that it received the addition -t (at a period when *au had already become ǭ > o [emphasis added]) by association with *et “out, out of” (Quendi and Eldar, circa 1960, WJ/367).

At the time of its making nāba-grota had no doubt already reached its archaic S form nǭv-ʒrot > Novrod. Grodnof has the same elements in the later more usual Sindarin order. The form Nogrod which later became usual is due to the substitution of Nog-, taken as a form of Naug “dwarf” (with the usual change of au > o [emphasis added]), after the element Nov- has become obscure (ibid., WJ/414).

... in Sindarin, in which P.E. *au and *ā had both become õ (open as in English pronunciation of au), and then > au, which however only remained in stressed monosyllables, and otherwise again reverted to o (NM/367-8, note #18).

There are numerous examples of this phonetic development:

• ᴹ✶tháurēnā > N. thoren “*fenced” (Ety/THUR).
• ᴹ✶ábārō̆ > ON. abōro > N. Afor “Refuser” (Ety/AB).
• ᴹ✶Phayanāro > ON. Phayanṓr > N. Feanor “Radiant Sun” (Ety/PHAY).
• ᴹ✶Anār > Anaur > N. Anor “Sun” (SD/302).
• ✶aklā́ra > aklōra > aglaur > ᴸN. aglor “brilliant, glorious” (PE18/87).
• ✶akjāwē-n > agauwen > agowen, past tense of S. caw- “*to taste” (PE22/152).
• ✶findarātō > find-raud > findrod > S. Finrod (VT41/9).
• Kh. Khazād > chaðǭd > chaðaud > S. Hadhod “Dwarf” (WJ/388).
• ✶lāmina > lǭ́m̃en > S. loven “echoing” (PE17/133).

The mechanisms of this sound change are somewhat unclear. In both Sindarin and Noldorin the diphthong [au] could arise from a variety of sources: the primitive long vowels [ā] and [ǭ], and the diphthongs [au] and [ou]. In Noldorin primitive [ou] became [au], but in Sindarin primitive [ou] became [ū], and only later diphthongs [ou] that arose from the vocalization of spirants developed into [au] (PE17/99). Furthermore, in Noldorin it seems that all primitive forms passed through [ǭ] on their way to becoming [au], as indicated by Tolkien’s chart of diphthongal developments from the Comparative Tables from the 1930s (PE19/25), though the development of primitive [ou] in this chart was not consistent with the phonetic developments appearing in The Etymologies.

Given this, it is very tempting to assume that the long slack [ǭ] never actually became [au] in polysyllables, but rather shortened and developed directly to [ŏ]. This is partially supported in Noldorin by the fact that Tolkien often represent [ǭ] as ō in The Etymologies, and the actual sound might have become [ō] before developing into [ŏ] or [au]. I think it’s plausible that in the Noldorin of the 1930s at least, [ǭ] only survived in monosyllables, and as part of becoming an overlong vowel it shifted to [au]; in polysyllables, however, it simply reduced [ǭ] > [ō] > [ŏ] and never became [au] at all.

As neat at this phonetic development is, there are a fair number counterexamples against it in Sindarin (see above), and even a Noldorin example against it from the late 1940s: ᴹ✶Anār > Anaur > N. Anor “Sun” (SD/302). These examples clearly show [ā] (> [ǭ]) > [au] > [ŏ], so that the sound definitely become [au] before its reduction to [ŏ] in Sindarin polysyllables. Furthermore, unlike Noldorin, there is no evidence that [au] became [ǭ] at any point in its normal development (e.g. in monosyllables), so that the tidiness of the Noldorin development is already broken by differences in how primitive diphthongs developed in Sindarin. Nevertheless, in some of his notes Tolkien indicated that [au] passed back through [ǭ] on its way to reducing to [ŏ] in polysyllables (WJ/367, NM/367 note #18), so that the full Sindarin development of [ā] in polysyllables is: [ā] > [ǭ] > [au] > [ǭ] > [ŏ].

Note that aw ([aw] or [au̯]) generally survives in polysyllables when followed by other vowels: S. and N. glawar “gold (light)” (PE17/159; VT41/10; Ety/LÁWAR), S. and N. tawar “forest; wood” (UT/319; Ety/TÁWAR), S. tawen “wood (of material)” (PE17/115) or tawaren (Ety/TÁWAR). See the entry on how [awa] sometimes became [au] for further discussion.

Retention of au: The sound change [au] > [ŏ] is very widespread, and seems to remain an active phonetic rule in both Noldorin and Sindarin, so it would apply to new compounds as well as old. However, there are a fair number of polysyllables that have au in both Noldorin and Sindarin:

• S. Anfauglir “Jaws of Thirst” (S/180).
• S. and N. (An)fauglith “Gasping Dust, Thirsty Sand” (S/150; Ety/PHAU).
• S. and N. Bauglir “Constrainer” (S/104; LR/206).
• S. and N. Draugluin “*Blue (Were)wolf” (S/174; LR/134).
• S. and N. Gaurhoth “Werewolves” (S/156; LR/284).
• S. Gaurwaith “Wolf-men” (UT/85).
• S. Glaurung “Gold-worm” (S/116; UT/75) vs. N. Glómund (LR/255).
• S. Gorthaur, a name of Sauron (S/32).
• S. ilaurui “*daily” (VT44/28).
• S. Naugrim “Dwarves” (S/91), N. Nauglar (LR/405).
• S. Nauglamír “Necklace of the Dwarves” (S/114), N. Nauglavir (Ety/NAUK).
• S. and N. Rauros “Roaring Spray” (LotR/366; RC/327; TI/285).
• S. Rhudaur (LotR/1039).
• S. Tauron “Lord of Forests” (S/29), N. Tauros (LR/206).
• N. Taurost “High City” (WR/260).

In some of these examples, the presence of an existing short o in one syllable of the word may have prevented the reduction of au in another syllable via dissimilation: Gaurhoth, Gorthaur, Rauros. The same might also be true of u in Glaurung and Rhudaur. Other examples are hard to explain, though: aur “day” remained in ilaurui but reduced in names of the week like Orgilion and Ormenel; naug “dwarf” remained in Naugrim and Nauglamír but reduced in Nogoth and Nogrod; glaur “gold” remained in Glaurung but reduced in Glorfindel, Glóredhel and Rathlóriel “Golden-bed”.

@@@ also preserved in Araw and aw past tense of sav-

Simplification to ó: The last two examples Glóredhel and Rathlóriel are of particular interest, because they are examples of cases where au did not reduce all the way to short [ŏ], but simplified to long [ō]. Similar examples include:

• S. and N. Anórien “Sunlending” (LotR/750; RC/776).
• N. Ardórin “?Day of Kings” vs. ᴹQ. Tárinar (PM/130), perhaps containing taur³ “king”.
• N. Glómund, earlier name for Glaurung (LR/255).
• S. and N. Nórui “June, *Fiery” vs. Q. Nárië (LotR/1110).
• S. Ódhel “Elf who left for Aman” from ✶aw(a)delo (WJ/364).
• S. Ered Lómin “Echoing Mountains” and Dor-lómin (S/118).

The last two names are outliers, because lómin was designated a North Sindarin word in the 1950s (PE17/133) or a loan word from Ilkorin in the 1930s (Ety/LAM), and both were likely attempts to retain early Gnomish names that Tolkien happened to like the shape of (G. Dor Lómin). In the other examples, however, it is probably notable that the long ó was retained in a stressed position. Perhaps the reduction all the way to ŏ was not required if stressed long ó was followed only by a single consonant.

Rare au > a: @@@ all from primitive ā before clusters: Narbeleth, Nardol, Nardol, Narwain (from [nār-]), Aracchír (from [arāt-]) and Iarwain (from [jārā-]) vs. Iorhael, Rodwen, Rodnor, Novrod, Arodreth, condir (from Discord conversation with Elaran on 2019-07-08).

Possible phonetic rules: Sorting through all these variations and coming up with a coherent set of phonetic rules is challenging, especially since we don’t know which forms represent conceptual variations or cases where Tolkien simply chose not to update names from an earlier phase of the language. Here is my attempt, however:

• If unstressed (e.g. in the final syllable of a polysyllable) au almost always reduced to short ŏ. The only exceptions were when there was an existing o or u in another syllable, where the reduction was inhibited: Gorthaur, Rhudaur.
• When stressed (e.g. in the initial syllable of most polysyllables) au sometimes only reduced to long ō (especially if followed by a single consonant) but often was reduced to ŏ (particularly if followed by consonant clusters). On other occasions au was retained, especially as the initial element in recognized compounds or when reduction was inhibited by an existing o or u in another syllable.

Note that the reduction of au > o could rarely occur even in monosyllables, such as the preposition o “from” derived from ✶ăwă or ✶au(t) (PE17/148, WJ/366) or do a (mutated) variant of taw “thither” (PE19/104). Minor words like prepositions were often unstressed and practically proclitics in Sindarin, since they mutate the initial consonant of following words, so these reductions might be the result of treating the preposition as part of the following word.

[ae] rarely reduced to [ĕ]: There are a few examples where it seems [ae] likewise reduced to short [ĕ] in both Sindarin and Noldorin:

• ✶ninkwiraite > ni(m)phraed > S. niphred “pallor” (PE17/55).
• ᴹ✶NÁYAK [> *naegro] > N. negro “to pain” (Ety/NÁYAK).
• Mewrim “*gulls” (WJ/190), likely the class plural of N./S. maew “gull”.
• athelas “kingsfoil” = athae + lass (PE17/49).

This change seems to be the exception rather than the rule, however, since there are far more examples where ae is retained in polysyllables for both Sindarin and Noldorin. Another possible example is S. Tengyl “Signifer” which appears beside another form Taengyl (MR/385), but as pointed out by Lokyt in a post on the Aglardh forms on 2019-03-16, very likely these two have different primitive forms, being cognates of the variant Quenya forms Tannacolli and Tainacolli respectively.

Conceptual Developments: There are examples of au > o in the Gnomish of the 1910s and Early Noldorin of the 1920s, as pointed out by Roman Rausch in his Historical Phonology of Goldogrin (HPG/§1.2) and Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.2.5). This indicates that this sound change was probably an old idea of Tolkien’s:

• G. Solmoth “Lord of Winds” from earlier Saulmoth (GL/68).
• G. and ᴱN. Balrog containing graug “demon” (GL/21, 42; PE13/138).
• ᴱN. arog “swift” vs. ᴱQ. arauka (PE13/137, 160).
• ᴱN. turhod “throne” containing ᴱN. haud “seat” (PE13/155).

Rausch pointed out that in Early Noldorin at least, the clear examples of au > o all appear in final syllables (HPITN/§4.2.5), which is consistent with the idea proposed above that this reduction was tied to patterns of stress.

GS/§4.168 @@@