OS. long final vowels were shortened; [-SV̄] > [-SV̆]
Primitive Elvish words often ended in long vowels, but in (Old) Sindarin these vowel were shortened in polysyllables, and later still, these short final vowels vanished. The evidence for these sound changes is so ubiquitous that it is not a question of whether these changes occurred, but when.
In point of fact, this reduction of primitive long final vowels to short vowels occurred not just in Sindarin, but in all Elvish languages, and at every conceptual stage. There is evidence that this change could not have occurred in Primitive Elvish itself, however, since the long final vowels persisted for varying lengths of time in different child languages. For example in Quenya, the accusative case was declined by lengthening the final vowel up through Classical Quenya (vanishing only in Tarquesta or “Modern Quenya”), indicating that long final vowels survived for quite some time in the Quenya branch.
There are examples that indicate a few Sindarin-specific phonetic developments took place before the shortening of long final vowels:
- final i-diphthong became long [ī] in polysyllables: ✶Ossai > ossī > ussi > S. †yssı̯ “Ossë” (archaic), later Gaerys (WJ/400).
- medial [s] became [x] before [j], [w]: ✶lisyā > liχı̯ā > leχı̯ > leich > S. laich “sweet” (PE17/148).
Nevertheless, it is probable that this final-vowel-shortening took place very early. At the minimum, this phonetic development probably took place before the various sound changes associated with long vowels in Sindarin, since there is no evidence such long-vowel sound changes applied to final vowels.
Before [ē] > [ī]: It is very probable that final-vowel-shortening took place before [ē], [ō] became [ī], [ū] in Old Sindarin. Otherwise, the resulting final -ī would have caused i-affection in such words, but there are plenty of examples where this did not happen, such as ✶ornē > S. orn “tree” and not **yrn (Let/426), or ✶kwenedē > S. †penedh “Elf” and not **penidh (PE17/140).
Before [ā] > [ǭ]: It is very probable that final-vowel-shortening took place before [ā] became [ǭ] in Old Sindarin. If it had, the well known phenomenon of a-affection could not have occurred, but the are plenty of cases where it did: ✶dirnā > S. dern “tough” (PE17/154); ✶russā > S. ross “red-haired, copper-coloured” (VT41/10).
An interesting related development is the formation of the Sindarin/Noldorin imperative, which is inflected by adding -o to the end of the verb. Since the primitive imperative particle was ✶ā (WJ/365, PE22/139), the likely origin of this Sindarin inflection was [-ā] > [-ǭ] > [-au] > [-o], since [au] became [o] in polysyllables. At first glance this may seem to contradict the analysis above. However, the imperative particle ✶ā was an independent element in Primitive Elvish, which could appear either before or after the word it modified (PE22/139-140). It seems likely that the independent [ā] first evolved to [ǭ] or even [au] before it was attached to the verb in Sindarin, and only thereafter did it evolve following ordinary phonetic rules, ultimately becoming [-o]. This is hinted at in the phonetic development of the imperative S. avgaro “don’t do it”:
Other later developments: There are a number of examples that point to various other phonetic developments occurring after final-vowel-shortening, in both Sindarin and Noldorin:
- voiceless stops aspirated after consonants except [s]: ✶mbartā > ṃbartā > ambarta [> ambartʰa] > S. amarth “fate” (PE17/104).
- intervocalic [s] became [h]: ᴹ✶barasā́ > barasa > ON. baraha > N. bara “fiery, eager” (Ety/BARAS).
- [oi] became [ui]: ᴹ✶wāyā > wōia > ON. uia > N. ui “envelope” (Ety/WAY).
There are additional Noldorin examples that show final-vowel-shortening before certain Noldorin-only sound changes, that nevertheless can give an idea of the history of Sindarin phonetic developments based on the timing of analogous Sindarin sound changes.
- [z] became [i] before [d] and [g]: ᴹ✶Ezdē > Ezde > Eide > ON. Īde > N. Îdh “rest” (Ety/EZDĒ).
- medial [s] assimilated to following liquid: ᴹ✶gaisrā > gǣsra > ON. gērrha > N. gaer “dreadful” Ety/GÁYAS).
- initial [s] became [θ] before [l], [r]: ᴹ✶slaiwā > slaiwa > ON. thlaiwa > N. flaew “sick(ly), ill” (Ety/SLIW).
Possible Late Survival: There is one Sindarin example that hints at a much later survival of long final vowels:
This appears to be a restoration of ᴱN. caifr n. “flea”, which was also derived from ᴱ✶kamp(r)u. If this derivation is accurate, it indicates that long final vowels survived beyond the spirantalization of aspirates. I think though, that there are other peculiarities with this derivation. For example, the primitive û might represent an over-long final [ū], which could have influenced the survival of the vowel’s length. For now, I am assuming the final-vowel-shortening happened earlier than this.
Conceptual Development: As noted above, there are no long final vowels at any stage of Sindarin’s conceptual history, and thus this sound change applied equally to Sindarin, Noldorin and Gnomish, though it is possible Tolkien adjusted the exact timing of final-vowel-shortening.