Is it because of the Babylonian exile that Deutero-Isaiah tempered the expection of Proto-Isaiah, from "death will be swallowed up" to "death, but after a long, full life"?
The Isaian apocalypse (ch. 24-27) usually is dated to the exilic or post-exilic period; the swallowing up of death is likely a restoration metaphor, as it is in Ezekiel 37. The exile is mentioned in 27:8-9, and the preceding chapter compares the return of nation through the gates of Jerusalem with the resurrection of the dead (26:1-2, 19). The exile is probably alluded to in Isaiah 25:8: "He will swallow up death forever. Yahweh my Lord will wipe away the tears from all faces; he will remove his people's disgrace from all the earth". This section is also interesting for its utilization of Canaanite mythological traditions: Yahweh slaying the dragon Leviathan (27:1), his celebatory banquet atop his mountainous abode (25:6-7), and the defeat of death (v. 8); Yahweh swallowing up death is a reversal of the story of Mot (Death) swallowing Baal. The myth of Baal's descent to Sheol and then redemption offers a useful parallel to the situation of the exile and restoration.
The eschatological picture in Trito-Isaiah was developed further in Enochic Judaism, which extends to everyone the longevity and health of the antediluvian patriarchs (1 Enoch 5:8-9, 10:16-18, 24:4-6; Jubilees 23:24-29).
Also, it seems that another factor against the Society's "translation" of Isaiah 65:20 is that Hebrew parallelism is being employed, with "dies at a hundred years" being parallel to "falls short of a hundred" (so the NRSV). The NWT destroys this paralelism. Your thoughts?
That's a good point. I would also point out that the three preceding colons have subjects characterized by their age: (1) `ûl "infant", (2) zaqen "old man", (3) ben-me'ah šanah "one who reaches a hundred years" (lit. "son of a hundred years"). So parallelism would lead one to prefer the fourth subject to be defined by age as well: (4) hachôte' ben-me'ah šanah "one who fails to reach a hundred years". "Sinners" doesn't fit this pattern at all. And the argument for parallelism is even stronger since both clauses have the same phrase (ben-me'ah šanah) in the subject.
The mistranslation in the NWT has a pedigree in the KJV and RSV, which ultimately depends on the Vulgate and LXX renderings (hachôte' "one who fails" rendered as hamartòlos "sinner" in the LXX). The other Greek versions (Aquila, Theodotion, Symmachus) instead render the word in question as hamartanòn "those who miss the mark, those who fail" (a form of hamartanein).