X-Digest-Num: 101
Message-ID: <44114.101.561.959273824@eGroups.com>
Date: Wed, 31 Mar 1999 20:20:00 -0800
From: Clark & Janiece Nelson <cjnelson@cybcon.com>
Subject: Tagged termsets and errata
X-Yahoo-Message-Num: 561
Content-Length: 1614
Lines: 41

Jorge J. Llambías wrote:
> >3- How can I say "A few years later"?
> 
> I would say {baza loi nanca} or {baza lo nanca be li so'u}. The
> term tagged with ZA would be the magnitude of the time offset.
> The book has a different method, using termsets. It's explained
> in pages 250-251.

Well, yes, so it is. According to le cukta, "It is grammatical for a
termset to be placed after a tense or modal tag rather than a sumti,
...."

Unfortunately, the grammar(s) in the very same book say otherwise. In a
term, a tag (or FA) can precede either a simple sumti or a KU. A term
can also be an unadorned termset. It can't be a tagged termset.

Since le cukta is self-contradictory, this qualifies as an erratum.
Several interesting questions suggest themselves:

1. Which way should the conflict be resolved? I remember some of the
discussions pointing out the need for a facility like this, and I
acknowledge the lack without it. On the other hand ... well, it's rather
unfortunate to have to modify the grammar after the baseline.

2. Is there a compendium of errata on some web site somewhere? I
remember there being a few Official Errata Announcements on the list
quite a while back, but either they're not collected on the Web or
they're hidden.

3. Is there an ETA for the baseline parser? The lack of it made
impossible the obvious quality assurance step of confirming the
grammaticality of all the examples. For instance, both the examples in
the section under discussion use "lu'a" for "la'u". That's a nit, but it
would be nice to be sure there aren't other nits lying around.

--
Clark Nelson