refer-person

[Marketa-Lopatkova]

If I am correct, the current version identifies the refer-person attribute:

• based on the person grammateme
• based on the morph. tag - whenever the person grammateme is not available but there is the relevant tag position)
  As a consequence, it appears also for some verbs (e.g., for být in ln94210_111.umr, r. 41, or počítat-006, r. 109) .

The refer-person attribute is appropriate only for personal pronouns?
This would mean for:

• nodes with sempos n.pron. def.pers.* OR n.pron.indef
• probably also for newly created entities?

[choroba]

It seems gram/sempos is often incorrect, e.g. the form "tu" can be adv.pron.def even when the morphological lemma and tag are ten and PDFS4----------, respectively. The morphological annotation was wrong in PDT-C 1.0, but was later fixed, while the sempos was not updated.

[Marketa-Lopatkova]

Personal pronouns (UMR person) in coreferential relations inherit their person value from the antecedent (if I am right). This does not work in case of a dialogue of 2 persons - then, the person changes in each reply (1->2, 2->1).