Augment (was Re: German ge- ptcpl cognates?)

[Vidhyanath Rao]

[ moderator re-formatted ]

----- Original Message -----
From: "Miguel Carrasquer Vidal" <mcv at wxs.nl>
Sent: Wednesday, February 02, 2000 6:43 PM

> "Vidhyanath Rao" <rao.3 at osu.edu> wrote:

>> There is an interesting typological problem here. According to Bybee et al
>> (The evolution of grammar), unmarked past vs marked present is unknown in
>> extant languages. This makes the usual classification of forms in Hittite
>> (and PIE) quite unusual. I remember asking about this before. Miguel
>> suggested Akkadian as another such example, quoting Lipinski to argue that
>> iprus was preterite, iparras was present. But in `Outline', Lipinski
>> explicitely assigns iparras to imperfective (putting present-future in
>> quotation marks). So the anamoly still unexplained.

> Still, the unmarked form is a simple past, while the marked forms
> are the imperfective ("durative", "present-future") with
> geminated C2, and the perfect (CtCC [iptaras], with infix -t-).
> Such a system is potentially very close to one with unmarked past
> vs. marked present (all it takes is the loss of the perfect).

Is it s a simple past or narrative past? [zero forms do survive as
subsequent forms even when they have been ousted from isolated
sentences, conversation etc.]

First: If something was imperfective rather than present/past, it had a
role in past imperfective. So the loss of perfect(ive) can only lead to
a perfective limited to past vs an imperfective. For what you propose
happened, the imperfective must have split into a past imperfective and
non-past imperfective and both the perfect and past imperfective must be
lost while the non-past imperfective survived. Show me an unmistakable
example.

Secondly. the seeming simplicity does not take into account the fact
that losses are not random, but display definite preferences. Perfect
tends to oust past/perfective rather than the other way around. More
clearly marked forms oust the unmarked forms rather than the other way
around. [Both of these, I thought, were old and generally accepted.
Didn't Kurylowicz put this in his methodological chapters in one of his
books?]

Isn't the whole point of typological studies that we should propose more
likely alternatives over positing a string of unlikely events?