In Quenya all primitive long final vowels shortened, excluding monosyllables and handful of other rare exceptions such as the strong imperative Q. avá “don’t!” (PE17/143; WJ/371). Unlike the corresponding sound change in Sindarin, this phonetic development was a continual process that was still ongoing quite late in Quenya’s history, being a factor in either late Parmaquesta [PQ] or early Tarquesta [TQ]. In notes on the Namárie poem written in the 1960s, Tolkien lists the absence of long final vowels as one of the signs that this poem was in TQ:

Galadriel’s song is in Quenya “Tarquesta”, that is the “colloquial” form of the language, though with some archaisms (dual) and poetic words, and abnormal poetic metric word order. The signs of “lateness” are ... the shortening of all vowels in final syllables (PE17/76).

As further evidence that long vowels still survived in PQ: the Quenya accusative remained an active part of PQ and this inflection was represented with long final vowels (Plotz). The accusative case was lost in TQ (PE17/76). However, the accusatives are a bit of special cases, because those originated from ancient over-long (trimoric) vowels (PE21/76-77), and thus may have shortened later than long vowels in general.

Thus, not all long vowels were lost at the same time. For example, in ancient compounds it seems long final vowels were lost at a comparatively ancient period, before the Quenya stress shift:

The degree of stress was, however, sufficient in all branches at an early period to cause certain vocalic reductions in (anciently) unstressed syllables, which are common to the Eldarin tongues (though not necessarily developed before their actual separation) ... Also often interiorly in long words with a far-retracted accent — for the most part in old compounds — ... long vowels were often reduced to short vowels in medial or final syllables (Outline of Phonetic Development [OP1], 1940s, PE19/57).

Furthermore, the resulting short final a, e, o were lost in long compounds, something not true of shorter words (PE19/59).

It seems long vowels shortened generally in final syllables, not just final vowels. Again, there are rare exceptions such as Q. palantír. Examples of long-vowel reduction in final syllables include:

• Val. akašān > Q. axan “law, rule, commandment” (WJ/399).
• ✶mbelekōro > Q. Melkor (WJ/402).
• ᴹ✶Bắlī + ndṓrḗ > Bálīndṑrē > Bálĭndṑrĕ > Balindor > ᴹQ. Valinor (PE19/59).

This last example is a strong argument that long vowels were shortened first in long compounds, since we see -ōrē > -ōrĕ [> *-ōr(ǝ)] > -or. Thus the shortening of long vowels in final syllables was in two or possibly three waves: (1) final vowels at the end of long compounds and (2) vowels in final syllables, including syllables that became final after short vowel losses and (3) original over-long trimoric vowels (like with the accusative) first became ordinary long vowels, and then became short vowels. Compare Q. Valinórë with Valinor.

Conceptual Development: Tolkien mentioned long-vowel-shortening in final syllables as far back as the Qenya Phonology of the 1910s:

Shortening of all final syllables in dissyllables & longer words by reduction of consonant groups and by the shortening of long vowels (PE12/4).

@@@ separate entry on early shortening in compounds; compare order with stress shift