pLab 6 Grading Rubric


Thanks for describing how you handled the extra parses
with modifiers!

It looks to me like -u marks not imperative, but irrealis,
and that irrealis is used in the imperative construction along
with other constructions.  Semantically, according to Ginzburg
and Sag, I think you'd find command_m_rels going with the
imperative use and outcome_m_rels going with at least some
of the other uses of the irrealis.  Accordingly, you probably
don't want to use imperative-clause with embedded clauses
with irrealis form verbs (as you'll get command_m_rel from it).


SENTENCE is probably redundant to MSG, but I think you'll
continue to want something analogous to IMPER and QUES
(but perhaps you can get away with a single feature FORM
that takes values such as irrealis, interrog, etc.).
This would canonically be a HEAD feature, though you might
find it more convenient to make it a CAT feature.