Lab 7 Grading Rubric


Even within the broad classes of scopal and intersective
adverbs, expect some subclasses with different MOD values.

"Intersectal" is odd, "intersective" being the usual form.

Rather than restrict the phrase type that scopal adverbs can attach
to, I would prefer a strategy of restricting the VAL inside MOD of
those adverbs.  I would be tempted to create a verbal 'ne', looking to
modify a [SUBJ cons, HEAD verb] constituent.  (Since there's a
reasonably good chance we'll need to do something semantically
different with constituent negation 'ne', this extra entry might not
be such a bad thing.)  The problem then becomes requiring the verb to
have not yet picked up any of its complements, either, though
modifiers are okay.  One strategy that ones to mind: use the features
LIGHT and HC-LIGHT (misnamed perhaps, for this use) so that both
lexical Vs and lexical Vs with modifiers (but no arguments) attached
are [LIGHT +].  This will probably require minor changes to the
constraints on HC-LIGHT in the Matrix.

I think in order to get all of the word orders, you might need to
go with an argument-composition analysis instead of a subject-raising
analysis.