In both Sindarin and Noldorin, any [ɣ] that survived other sound changes eventually vanished. At this point, the only surviving [ɣ] appeared primarily (a) between vowels or (b) at the end of words. Furthermore, these [ɣ] were mostly the result of the soft-mutation of [g]; any [ɣ] that originated from Primitive Elvish vanished much earlier. Where this [ɣ] appeared medially before and after sonants (liquids, nasals and other spirants), it had developed into vowel (usually i but sometimes u in Noldorin). Where [ɣ] became final after another consonant, it generally had already developed into the vowel a.

Tolkien explicitly mentioned the vanishing of [ɣ] from [g] in notes on the Noldorin usage of the Feanorian Alphabet from the 1930s:

Early in the third century the change (beginning to appear at the end of the second) was complete, whereby the voiced spirant [ʒ] derived from ON g became vocalized, and was lost between vowels. It was long retained in Gondolic spelling (f transcribed gh), especially where it had not become a syllabic vowel as in tara “tough” < targh, ON targa, and its original place was still marked by hiatus. But later in Gondolic spelling the use of f became rare. It was sometimes retained to mark the mutation of initial g: it is then transcribed gh or ” since no consonantal sound was actually preserved: thus snjn4 golodh “Gnome”, `Bfnjn4 igholodh or i”oloð “the Gnome”. The old h-sign ½ (no longer required as h since this function was exercised by 9) was reintroduced and called gas-dil “stop-gap”, and employed to mark hiatus caused by the loss of g, ʒ and so (since the medial-hiatus was usually got rid of by contraction) usually a sign of the mutation of g: as `B½.j.4 [sic] i’oloð “the Gnome”, gasdil is thus transcribed [’] (PE22/34).

As indicated by this note, the loss of [ɣ] from [g] played a significant role in the soft mutation system of Sindarin and Noldorin: whereas the soft mutation of the other voiced stops b and d were the voiced spirants v and dh, an initial g vanished completely, which was often marked with a [’] in transcription. There are numerous examples of this sound loss medially in Sindarin and Noldorin as well:

• ✶(ñ)galata-rigelle [> *galatariʒel(le)] > S. Galadriel (MR/182, 470; PE17/50, 60).
• ᴹ✶Tūgore > tūghor > N. Tuor (Ety/GOR, TUG).
• ✶magiti- [> *maʒit(e) > *maid] > S. maed “shapely” (VT41/10).

As the last example shows, the loss of [ɣ] from [g] was early enough that any resulting diphthongs underwent the same sound changes as other late diphthongs, [ai] > [ae] in this specific example. There is evidence that the loss of [ɣ] at morpheme boundaries behaved somewhat differently than it did in the interior of words. An interesting case study is the name Maglor. In notes from the last 1960s:

Makalaure was converted simply phonetically to S. maglaur > maglor. Its pure Sindarin [development] would have been maka-glaur-. [In] S. glaware > glawar = Q. laure but as second element in compound glaware > glaur. magalor- (VT41/10).

This example indicates that in normal Sindarin phonetic development, a [ɣ] at the beginning of a morpheme was simply lost (-glaur > -ʒlaur > -laur > -lor), even in cases where it would have undergone different developments medially: normally [aɣl] > [ail] > [ael], as in mael “well” < √MAGA [maglā] (PE17/162). The disappearance of [ɣ] at morpheme boundaries is also well-supported by the behavior of soft-mutation: the lenited forms of words beginning with gr- and gl- consistently show r- and l-. However, in the 1950s Tolkien sometimes wrote the name Maglor as Maelor, indicating he may have vacillated on the exact phonetic developments.

Conceptual Developments: Since the disappearance of initial g in lenited forms dates all the way back to Tolkien’s earliest notes on Gnomish Grammar from the 1910s (PE11/7), it seems this phonetic development was part of Tolkien’s conceptual framework of the languages from the very beginning. There are clear examples of this sound loss in both Gnomish of the 1910s and the Early Noldorin of the 1920s, as discussed by Roman Rausch in his Historical Phonology of Goldogrin (HPG/§2.6) and Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.1.3):

• ᴱ✶nēgittĕ > G. nith “wax” (GL/60).
• ᴱ✶Eka-ʒalda-mbod-t > ᴱN. Egallmoth “He of the Wide Shoulders” (PE13/142, 159).

However, there are a couple of examples in Early Noldorin where it seems [ɣ] sometimes vocalized to [i] between vowels instead, as also pointed out by Roman Rausch (HPITN/§4.1.3):

• ᴱN. poi “aft”, ᴱN. poiant “aft”, loan words from ᴱT. págas and ᴱT. págant (PE13/152).

The exact conditions whereby intervocalic [ɣ] became [i] in Early Noldorin are unclear.

GS/§4.170 GS/§4.171 GS/§4.172 @@@

HSG/§2.6 @@@