Yet another areal feature in SA languages?

[Vidhyanath Rao]

I had written
>Dahl gives the Japanese ``-te shimaru'' construction and the (modern)

Jonathan Silk wrote
JS> I think at least in modern Japanese this must be -te shimau.

You are right. This ought to teach me to keep the books I quote from in
front of me, but I never seem to learn.

---

Brigit Kellner wrote

BK> Regarding the Japanese -te shimau connection, it is to be noted that
BK> the completive aspect of this construction applies throughout *all*
BK> tenses and forms, and is not limited to the past.

The same is true of Tamil, and a few of the examples of Bybee et al. (I
checked only four of the 20+ languages they list, in some cases the
grammars are silent on this.) Of course in RV, the aorist stem is used
in the moods also, even if sigmatic optatives and imperatives are
extremely rare.

-----

JS> [Witzel] has also alluded, again silently, to another feature ...
JS> namely that it implies a negative result;

This clears up a small puzzle. Text of Dahl's questionaire, about
announcing the expected arrival of the king would be ``aracar vandu
viTTaar.'' The viDu construction is almost obligatory here and I
couldn't understand why this was not in the list of correspondences.

BK> The feature of undesirability [...] is not always present [in the
BK> past].

Now my confusion is back. Will the -te shimau construction be used to
translate ``The king has arrived!''?

-----

I wanted to study more examples of ``completives'' before I tackled
Sanskrit or IE. But perhaps I need to explain in more detail my
motivation.

JS> I do not think you would have had to have sent the letter within
JS> that day, for example, as I understand the distinction in Vedic verb
JS> forms to suggest.

The problem is that Vedic aorist is not limited to events of the same
day. In fact, even in Paninean grammar, there are exceptions. Firstly,
the aorist is the ``past of the most recent X'', with day being just the
default value [3.3.135]. But this is minor quibble. There are living
languages which are reported to do similar things.

Secondly, Panini says ``lu"n; anadyatane la"n.'' According to Katyayana,
anadyatana is a bahuvriihi; so one can only say ``adya hyo 'bhukSmahi.''
(cf RV 8.99.1, 8.66.1). [But Patanjali and latter authorities seem to
contradict this by identifying na+adyatane with anadyatane and the
complement of bhuutamatre.] Thirdly, according Patanjali, the aorist can
be used when the speaker does not have the intention (vivakSA) of
indicating the time. This comes close to the ``aorist of statement''
(Konstatierung in Hoffman).

The question is if these are enough. Jan Gonda (The aspectual functions
of the Rgvedic present and aorist) maintained that the answer is no. He
gives a long list of aorists that occur with imperfects. He also
gives an example of imperfect used for event of the same day. The latter
does not really matter: Remoteness distinctions, in some languages that
have them, is not usually applied in narrative contexts. Or it might be
relative to the point of time in the narrative. Also, even closely related
dialects may differ on whether the distinctions are strict and
obligatory or based on the subjective conception of the speaker.

But the first one is less easily disposed off. What struck me is that
the Tamil -viDu construction has many of the properties that Gonda notes
in the RV aorist. Also, the simple past need not indicate completion: To
quote Annamalai's example ``DaakTarukku paNam koDutten. aanaal avar
vaangik koLLavillai.'' `I offered the doctor the money. But he did not
take it.' [Of course, due to the presence of a progressive such examples
are rare.] But it would be gross distortion to say simple past :-viDu is
the same as imperfective : perfective.

---------

MW> * imperfective aspect ( expressed by present stem, e.g. imperfect:
MW> a-ga-ccha-t; type of verbs: durative) and
MW> * perfective aspect (expressed by aorist stem, a-gan-(t), a-gam-a-t
MW> etc., not just the sigmatic aorist : a-gaM-s-/ agam-iS-)
MW> ... (e.g., K.Hoffmann, Der Injunktiv im Veda, Wiesbaden 1969).
MW> Both have diff. meanings, obviously, and should be translated (Vedic
MW> exx.) as: "he came :: he has come just now." Panini knows that.

Remoteness distinctions have nothing to do with aspect. In languages
that have both, remoteness distinction may be limited to the perfective
[Bybee, p.101] or interact in possibly complicated ways. I have not heard
of an example in which remote past = imperfective and recent past =
perfective.

I don't think that Hoffmann says that the RV imperfect is an
imperfective. He suggests a stage between PIE and Vedic in which the
imperfect is an imperfective limited to remote past and narration while
aorist is perfective in those context and is also used for recent past
and statements (without regard to aspect?). Without a living example,
this sounds too ad-hoc.

There is also a reason why I don't want to assume a priori that the root
and sigmatic forms were used suppletively from the beginning: There is
evidence to support the view that the distinction between the imperfect
and aorist is fuzzy in RV [Elizerenkova; Gonda `concurs' with this while
asserting an aspectual distinction!], and this has been arrtibuted
[Watkins?] to an original unmarked status of root forms. I don't think
that it is accidental that classifying root forms of RV sometimes gets
quite complicated.

MW> What about Russian (found in every grammar), or even the English
MW> forms in -ing ( I was sitting :: I sat)??? (and colloquial N.W. German).

English -ing is a progressive. In example 1 above, we can't say `I was
giving the doctor the money, but he wouldn't take it.'

Slavic distinction in curious. It denotes attainment of a limit, but
like perfectives of other languages, it is the usual form in narration.
To put it a bit differently, Slavic perfective is like the typical
perfectives in narrative contexts, but different in non-narrative
contexts [see Dahl, pp. 84--89, and Bybee et al. pp. 87--95]. But
completives are not the usual form in narration, and behave differently
from perfectives in all contexts. In fact it is possible to a language
to have a completive in addition to a perfective.

MW> I think one of our Amer. linguistic colleagues (who?) has written
MW> about this soem 5 years ago , as found in medieval Sanskrit
MW> (Narrative literature: periphrastic constructions)

I will appreciate ii if any one can give further pointers to this.

-Nath