Aspect/modstr/temporal annotation for stative verbs and other concepts with (abstract) rolesets

[Marketa-Lopatkova]

• NO difference between stative and action verbs ---> aspect/modstr/temporal annotation (preliminary)!!

• nine abstract predicates (as definedin the Guidelines) ... we use them only if we are sure that it is a predication --> aspect/modstr/temporal annotation (preliminary)!!

• other concepts with abstract rolesets (reification, implicit rolesets) ... the same as for pother -91 "predicates" --> aspect/modstr/temporal annotation (preliminary)!!

• have-quant-91, have-degree-91) ... we use them only if we are sure that it is a predication --> aspect/modstr/temporal annotation (preliminary)!!

• BUT: discourse relations ... NOT events (despite having rolesets)

• agentive nouns are annotated as verbs ---> events, i.e., with aspect/modstr/temporal annotation (preliminary)!!
  Šárka: a teacher teaches habitually vs. for a murderer, it is enough to commit just one murder to be a murderer.
  Do we want to mark the aspect at the verb to_teach, to_kill? YES!

• words denoting processes (válka, příchod, bouře) ---> events, i.e., use valency frame from PDT-Vallex or CREATE it (??where to store them?; NOT for automatic conversion) and annotate aspect/modstr/temporal annotation (preliminary)!!

[dan-zeman]

• BUT: discourse relations ... NOT events (despite heving rolesets)

The validator no longer treats discourse concepts as events (1cb51b6).

[dan-zeman]

@Marketa-Lopatkova you do not mention publication-91 as receiving special treatment but I think you said it would be exempt from the :aspect and :modal-strength requirement, too? Though in your excerpt from Julia's answers it sounds still undecided, or maybe leaning towards treating it as an event:

I think it's much more unusual for a lot of the other implicit rolesets to come up in a predicated form, but some of them do occasionally, and that goes for the -91 rolesets used for reification as well. (Generally, this is a copular or existential construction in English). I can't think of a case with publication-91 offhand-- maybe that one isn't one that occurs in English. My goal is to try to follow the same predicating/non-predicating guidelines with all of the -91 rolesets.

[Marketa-Lopatkova]

@dan-zeman you are right: publication-91 and other abstract rolesets for metadata information should be exempt from the :aspect and :modal-strength requirement as well. I've updated the instructions in the Events file.

[dan-zeman]

OK. As of 1b14b96#diff-5f3c94b4ce37d3d41d4d505c19f20bd288ed17c8ae575a988d802f8d6a8c5e10, publication-91 is not considered event and does not require :aspect and :modal-strength.

Should we go to the other extreme and disallow these roles with this concept?