gender aggrement

[Alexandre Rademaker]

A possible extension of a rule like

SETPARENT ("filho" @N<PRED) (1 ("de")) TO (*-1 (@SUBJ> PROP) BARRIER CLB-ORD) ;

is to make the target test using the lemma instead of the wordform "filho" and make it always agree with the gender of the <contextual_target>! How to make it? In that case, we would have a right analysis of cases for

X ... filho de ....
Y  .... filha de ...

[Tino Didriksen]

Dunno if you ever figured this out, but here are some solutions:

First, you could just write two rules, one asking for masculinum when target is filho, and another for femininum when target is filha. Assuming words are tagged Masc or Fem:
SETPARENT ("filho" @N<PRED) (1 ("de")) TO (*-1 (@SUBJ> PROP Masc) BARRIER CLB-ORD) ;
SETPARENT ("filha" @N<PRED) (1 ("de")) TO (*-1 (@SUBJ> PROP Fem) BARRIER CLB-ORD) ;

Second, you could use top-level set unification and group all masc/fem tags into sets and ask that they agree:
LIST Mascs = "filho" Masc ;
LIST Fems = filha"" Fem ;
SET Genders = Mascs OR Fems ;
SETPARENT &&Gender (@N<PRED) (1 ("de")) TO (*-1 &&Gender + (@SUBJ> PROP) BARRIER CLB-ORD) ;

See https://visl.sdu.dk/cg3/chunked/sets.html#set-unification-set for documentation.

-- Tino Didriksen