[matrix] Re: [rmrs] tense and mood info in rmrs
Anette Frank
frank at dfki.de
Thu Nov 10 10:45:26 CET 2005
Hi,
thanks to Lars, Dorothee and Melanie for their reactions!
I can see - and support - the concern for a common, cross-language
representation
for tense and aspect, though it seems there is not yet a clear picture
about what exactly
it should look like.
In my view, the best way to go forward towards this aim would be to come
up with a
collection of definitional criteria for the different features used in
the different grammars,
and gather reactions from the different language representatives, for
continuous refinement.
Besides the aim of a fine-grained and cross-linguistically appropriate
feature-set for tense
and aspect, we should, however, also consider the special concerns RMRS
was designed for
- to allow compatible and informative semantic representations from
shallow to deep parsers.
For a parsing component that does not have a very fine-grained lexicon
or rule sets for the
determination of proper tense and aspect features, it will be impossible
to deliver the
fine-grained tense and aspect representations that a deep grammar might
be able to.
However, a more shallow parsing component might well be able to deliver
precise
*morphological* tense and mood features that can be useful, as a
coarse-grained representation,
for post-processing components, even though they do not allow for the
very fine-gained
classifications.
It would therefore seem to me a viable solution to record morphological
tense features in the
RMRS, alongside the semantic tense and aspect classes, in order to
record basic tense
and mood distinctions from shallow parsers, where available. The
alternative would be to throw
away this information altogether, which doesn't seem reasonable to me.
Of course, also there,
one should see whether one could come up with a feature set that covers
as many languages
as possible.
Best, Anette
More information about the matrix
mailing list