OS. final i-diphthong became long [ī] in polysyllables; [-SVi] > [-Sī]
There is evidence that in polysyllables, any final diphthongs ending in i were reduced to a long ī in Sindarin, and this long vowel was later lost with various phonetic effects. This sound change has been noted by David Salo (GS/§4.33), and is needed to explain the development of Sindarin plurals, which regularly demonstrate i-affection and i-intrusion, such as S. erain plural of aran “king”. There are a few explicit examples of this plural development in Sindarin and Noldorin:
- -āi > -ī: ✶tekmāi > tekmī > tiŋw > S. tīw, plural of S. têw “letter” (PE17/43).
- -ōi > -ī: ✶góndōi > gondŏi > ON. gondī, plural of ON. gondo “stone” (PE21/58).
In theory this sound change might only apply to plurals, being a reformation of plural forms after the loss of final vowels in Sindarin. However, we also have an explicit example of this change for a word that is not a plural:
- -ai > -ī: ✶Ossai > ossī > ussi > S. †yssı̯, an archaic Sindarin cognate Q. Ossë, later called S. Gaerys (WJ/400).
This example (as well as the plural examples above) indicate the reduction of i-diphthongs was a general sound change, not specific to plurals, and probably occurred at an ancient stage of Sindarin’s history. However, some Telerin plurals indicate this changed did not occur in that branch of the language: Elloi, plural of T. Ello “Elf” (WJ/373); Lindai, plural of T. Linda “Singer” (WJ/382). Thus, this change probably did not date back to the Ancient Telerin period.
Many of these ancient i-diphthongs were long diphthongs such as -āi and -ōi. The likely mechanism for this change is that these first became short diphthongs -ăi/-ŏi (as indicated above), then the intial sound in the diphthong became less distinct [ǝ] and ultimately -ǝi > -ī. A hint of this final development occurs in a discussion of the tengwar representations of Old Noldorin from the 1940s:
The chief exceptions were [final] ī in standard ON (from ǝı), long and distinct from ĭ; also ū occurred finally distinct from ŭ (PE22/27).
It is an open question whether this sound change also applied to final u-diphthongs; this last note hints that it might, but there are no clear examples. Such final u-diphthongs would have been rare in Primitive Elvish, primarily the result of ancient dual forms. Since the dual was lost in Sindarin, even if the final u-diphthongs had a reduction similar to i-diphthongs, it would have had virtually no effect on Sindarin.
Conceptual Development: It is likely that Tolkien introduced this phonetic rule into Early Noldorin at the same time he introduced plural forms using i-affection, as early as the 1920s, though exactly when he formalized this rule is unclear.