Indo-Iranic languages have a much longer attestation, yes (if we include the loanwords in Mitanni, 1400 BC), but you should not equate that automatically with "older". We're talking about Late IE languages (where the so-called "pharyngeals") vanish with vowel-colouring qualities while in the Anatolian languages they were largely preserved. Further the Indo-Iranic branch is also satemized (like Armenian and Balto-Slavic). The Indo-Iranic languages have the distinct feature that they then merge *e, *o > *a (as well as the long counterparts, *ē, *ō > *ā - here I would like to ammend that the word for horse,
*(H)ekwos, is subject to this sound change, think of Persian "
asb" and Hindi "ašva" versus Latin "
equus"). However, at the same time (and this is where the limitations of the tree model come into play, where I agree with Alan, by the way), Iranic (but not Indic) shares with Armenian and Greek that word-initial *s- yields *h-. This shows you that we're talking about a late Indo-European dialect continuum. The latter issue, by the way, is also the reason why the Mitanni loanwords cannot come from an Iranian language.
Get factual. I would assume you're basing that idea of modern Kurdish, Hindi-Urdu and Pashto (most of all Kurdish, I presume, since that is a language you speak). However:
- the Nuristani languages have no ergativity.
- Sanskrit had no signs of ergativity.
- Pali had no signs of ergativity.
- Avestan had no ergativity.
- Old Persian had no ergativity.
- Ossetian has no ergativity.
-
Yaghnobi has no ergativity.
That's Bouckaert, Gray and Atkinson's work, we've discussed it in many threads here. Their methodology is dubious and their dates are bizarre. Note how many of their dates are completely nonsensical. For example Romani (the Gypsi languages) are supposed to have diverged from the other Indic languages in about 1500 BC, while in reality, the Romani people did not arrive in Europe before the Middle Ages.