You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: lib/Treex/Block/HamleDT/PragueDeprelsToUD.pm
+6-1Lines changed: 6 additions & 1 deletion
Original file line number
Diff line number
Diff line change
@@ -506,10 +506,15 @@ sub convert_deprels
506
506
# The AuxY deprel is used in various situations, see below.
507
507
elsif($depreleq'AuxY')
508
508
{
509
+
# An AuxY depending on an AuxC may signal a multiword subordinator, which could be converted to mark and later to fixed.
510
+
# However, some combinations are not fixed expressions. This could be decided using language-specific lists; another clue
511
+
# is that if the words are not adjacent, they probably do not form a compound subordinator.
512
+
###!!! See also HamleDT::CS::HarmonizePDTC::prevent_compound_subordinators();
509
513
# When it is attached to a subordinating conjunction (AuxC), the two form a multi-word subordinator.
510
514
# Index Thomisticus examples: ita quod (so that), etiam si (even if), quod quod (what is that), ac si (as if), et si (although)
511
515
###!!! But not always! E.g., we had an apposition "jako X, čili jako Y", it was hamledtized as a hypotactic structure, then "čili" ended up attached as AuxY to "jako", but they do not form a fixed compound subordinator!
###!!! Also not for double "že": "Myslí si, že když to máme za humny, že na to je dost času."
517
+
if($parent->wild()->{prague_deprel} eq'AuxC' && lc($node->form()) ne'čili' && $node->form() !~ m/^(čili|že)$/i) ###!!! raději bych měl testovat formu rodiče i dítěte, abych zase nevyhodil něco, co vyhodit nechci!
513
518
{
514
519
# The phrase builder will later transform it to MWE.
0 commit comments