[matrix] Re: [rmrs] tense and mood info in rmrs

[Melanie Siegel]

Good Morning,

the current Jacy Japanese HPSG contains a variation of aspect and mood:

ASPECT
progressive
prospective
inceptive
terminative
perfective
perfect_progressive
modal

MOOD:
indicative
imperative
negative
conditional
voluntative
consultative
potential
causative

I haven't yet quite thought about representation in RMRS (i.e., 
multilingual compatibility). But my feeling is as well that we shouldn't 
just add "perf" and "prog". I have documented aspect in my current 
documentation efforts, but not yet mood extensively enough. Kei 
Yoshimoto in his ph.d. thesis has written about Japanese and English 
aspect (connected to Verbmobil translation) and transfer rules for them.

So the question would be, if it is possible at all to come up with a 
representation that is useful for all languages involved. If we try, 
these should certainly not only be German and English.

Best,
Melanie


Dorothee Beermann wrote:

> At 19:53 04.11.05 +0000, you wrote:
> 
> Ann wrote:
> 
>> My opinion is that the tense and aspect representation is a matter for 
>> the
>> grammar writers and should be discussed on the Matrix list.
> 
> 
> and here from Annette's mail:
>  From the MRS representations produced in the ERG and the German
> grammar it would seem most straightforward to have two (or three)
> additional features:
> - boolean "perf" and "prog" to encode future/present/past perfect or
> progressive tenses.
> 
> As a comment to Annette's mail the following:
> 
> Progressive' can presumably be captured by 'PROTRACTED +, STATIVE -', 
> with the features already available. 'Perfect' means in part different 
> things, even in as closely related languages as German and English. Thus 
> it is not obvious that we would want to include 'perfect' in the 
> rmrs.dtd - at least if this is designed as a repository for multilingual 
> purposes.
> 
> An example of concern arises when we move beyond the closest European 
> languages,
> 
> When building matrix-grammars of Kwa languages, which are Aspect marking
> languages, we need to at least encode the following distinctions for 
> ASPECT and MOOD:
> 
> ASPECT:
> completive versus perfective
> habitual
> future
> progressive versus continuative
> consecutive  ( in serial verb constructions)
> 
> MOOD:
> indicative
> imperative
> 
> The distinctions given above (and more) are encoded in the 
> morpho-phonemics and for AKAN,
> for example, described in a paper by Osam:
> 
> http://edvarda.hf.ntnu.no/ling/tross/osam.pdf
> 
> Ways of  encoding the relevant distinctions is one of the major concerns 
> of the
> grammar writer when designing the verbal system of these languages and 
> thus an
> integral part of multi-lingual grammar engineering.
> 
> Even when faced with such an array of 'new' distinctions, we think the 
> right approach is to first try to construe these distinctions in terms 
> of the already (quite rich) available system. We thus second Ann in this 
> issue.
> (Of course, we also need a place where we can use 'received' 
> morphological labels of each language concerned - but, again, we don't 
> see that this would have to be in the rmrs.)
> 
> best
> Lars and Dorothee
> 

--