S. long voiceless spirants shortened; [θθ|xx] > [θ|x]
In both Sindarin and Noldorin, long voiceless spirants shortened, as noted by David Salo (GS/§4.184). Tolkien alluded to this phonetic development in a discussion of the transcriptions used by Ælfwine in his document on the Feanorian Alphabet from the 1930s:
The transcription used by Ælfwine, based mainly on Old English, and partly on Irish use, was variable ... These he represented thus: 3 by þ, ð (not initially) or th. Medially where Old English had not in his time a voiceless [þ] normally, he often used th, or even þþ, although the sound in Noldorin was never long in his time. e by f, ph. Medially where OE had not normally [f], he often used ph or ff, although [f] was never long in Noldorin in his time (PE22/033).
Based on this note, it seems all long spirants shortened in Noldorin: [θθ] > [θ], [xx] > [x], [ff] > [f]. There are examples of each of these phonetic developments in The Etymologies:
- ᴹ✶battā́ > ON. batthṓ > N. batho “trample” (Ety/BAT).
- ᴹ✶rokkō [> *roχχ(o)] > N. roch “horse” (EtyAC/ROK).
- ᴹ✶apparkā > ON. appharkha > N. afarch “arid, very dry” (EtyAC/A).
In Sindarin, however, it seems medial [ff] remained long:
PH ... in the middle of a few words where it represents a long ff (from pp) as in Ephel “outer fence” (LotR/1114).
The reduction of other long spirants was an aspect of Sindarin’s phonetic history, however. For example, S. roch had the same etymology as in Noldorin, and Tolkien explicitly described [θθ] > [θ] for Sindarin:
Of this already in Common Eldarin a diminutive/affectionate form was made with (as frequent in such words) reduplication of medial consonant > netthi. Of this Q. nette, T. nette, S. neth are normal developments (tth > Q., T., prehistoric S. tt; later S. þþ > þ) (VT47/14).
Conceptual Developments: Gnomish had the same reduction of long voiceless spirants, as noted by Roman Rausch in his Historical Phonology of Goldogrin (HGP/§2.4):
- ᴱ✶ı̯otta > G. gôtha- “possess, have, hold” (GL/42).
- G. hacha “the hams, buttocks” vs. ᴱQ. hakka (GL/47).
- G. laf “loose-end, end of rope, hem of robe” vs. ᴱQ. lappa (GL/52).
The same is true of Early Noldorin, as noted by Roman Rausch in his Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.1.5), though finding a complete set of examples is challenging: