--- In cybalist@..., Harald Hammarstrom <haha2581@...> wrote:
> [...] Are there are theories on how it developed?
> Independently in II and IA or joint? Due to some substratum?
> Why only in the preterite-ish tense?
The best theory I have seen is by Benveniste. [see the book with a
collection of linguistic articles, the name escapes me at the
moment.]
Basically the (Asian) languages with ergative in past tense are
'be' langauges (ie, languages in which possession is not
expressed by a verb like 'have' in English) in which an old
perfect ousted the older preterite (as happened in French more
recently, and seems to be quite common, Middle Irish for another
example). No substratum explanation is warranted and the development
is not common enough to be inherited. [In some Middle East iranian
languages (Sogdian?), there was a verb for 'have, namely 'dar' and
their past was dar+ta participle.]
In Indic, the development can be dated comparatively precisely. In
the oldest strata of the Pali canon (generally dated to about
250-200 BCE), the narrative uses the old preterite [amalgam of Old
Indian aorist and 'imperfect'] while direct speech is almost
invariably has the ta participle. But, the latter has ousted the
former in Prakrit lit datable to 100 CE. [Due to layers in the
Sanskrit Epics, the situation there cannot be taken to be older than
the earliest parts of the Plai canon. In fact, in the parts of
Ramayana that are considered to be older, direct speech has only
the ta-participle, while the narrative part has the 'imperfect' or
the reduplicated perfect.]
BTW, the ta-participle is in origin a resultative and the
construction indicates >present< state. [see Jamieson, the
tense of the predicated past participle, Indo-Iranian J, 198x (83?).
On the Pali situation, see 'Syntax of the infinite verb forms of
Pali' by Hendriksen]. It is quite natural for this be 'ergative'.
This construction was never a passive.