Open kimgerdes opened 4 years ago
There is no general rule. A sentence modifier must move to the head of sentence.
I am happy today In SUD : am -[mod]-> Today
A modifier of the adjective must stay on the adjective.
I am very happy In SUD : happy -[mod]-> very
The linguistically correct solution is to add lexical information to decide. An alternative is to enrich the UD annotation.
Remark : this issue is not put in its proper place. It needs to be transferred to : https://github.com/surfacesyntacticud/tools/issues
Yes, thanks, you are right. But wouldn't there be a (possibly idiosyncratic) heuristics that could be used by order or category?
And there is also the thing that when "happy" has an object, this is correctly kept on the adjective:
i'm happy that he came
idem for comp:pred
im happy to join
whereas
i'm happy about his arrival
is moved to the aux.
Actually i don't understand what the grammar does. For example I'm pleased with the results
keeps the PP on the ADJ. See http://match.grew.fr/?corpus=SUD_English-EWT@2.6&custom=5ee002cc65f33 Possibly only mod
is moved up?
Remark : this issue is not put in its proper place. It needs to be transferred to : https://github.com/surfacesyntacticud/tools/issues Yes, that would have been better, but it's also about discussing what structure we want (and want to describe in the guidelines) before considering that this is a bug in the translation grammar.
When a complement is an argument of an adjective as in the examples given by @kimgerdes, the conversion between comp:obl
or comp:obj
and ccomp
, xcomp
or obl:arg
works perfectly.
In the example i'm happy about his arrival
, we can consider the complement of the adjective rather as an argument than as a modifier.
The problem arises when the complement is a modifier. There is no way to distinguish a sentence modifier from an adjective modifier, except by using lexical information.
Can I close this issue ?
Decision: modifiers of predicative constructions in the form ((VERB | AUX )+ ADJ)
are attached to the verb if they are sentence modifiers (temporal, locative, judgment ... modifiers), otherwise they arre attached to the adjective.
Correction of the treebanks:
I think we have to modify the UD-SUD translation in case the adjective has dependents: http://match.grew.fr/?corpus=SUD_English-EWT@2.6&custom=5edeabd6e5e98 the dependents should to stay on the adjective, don't they?