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

[Ann Copestake]

The ideal would be that the shallow parsers come up with an underspecified 
version of what the deep parsers do, which should be viable if we factorise 
the information correctly.   (I think there's still some wiggle room in 
exactly what we mean by underspecification for features and values, if there's 
a formal issue.)

It may not be possible with the current feature set, but could we establish 
some examples that demonstrate the problem?  I still don't know why 
progressive doesn't correspond to some settings of the existing features and 
why you need a new feature.

Ann