search

UniversalConceptualCognitiveAnnotation / docs

UCCA Documentation
https://universalconceptualcognitiveannotation.github.io/
10 stars 1 forks source link

Secondary change-of-state/causative verbs supporting stative predicates #50

Open nschneid opened 5 years ago

nschneid commented 5 years ago

Currently, generic change-of-state verbs like "become" are treated as secondary, with the primary predicate as S: image

Likewise for causatives: image

This is proving problematic for SNACS integration, because the overall event is a change-of-state, and thus should be a P. It also is problematic for paraphrasing: the aspectual type for "Mary got pregnant" (D, S) would differ from its paraphrase "Mary was impregnated" (P).

One solution that keeps these aspectual verbs as secondary Ds would be to annotate D and S within a P unit:

Another would be to use multiple categories, e.g.:

"P+D" could be interpreted as "introduces a change of state, but secondary to another predicate".

(Contrast resultatives like "hammered the metal flat", which have two primary scene-evokers, so both P and S are represented.)

omriabnd commented 5 years ago

I think this is a great idea for a refinement layer. I feel that the foundational layer is too expansive for an initial coarse-grained layer as it is. Also note, that Ds inevitably change in translations and paraphrases. Lexicalization patterns may change a causative+verb to just a verb etc.

On Thu, Nov 29, 2018 at 8:39 AM Nathan Schneider notifications@github.com wrote:

Currently, generic change-of-state verbs like "become" are treated as secondary, with the primary predicate as S: [image: image] https://user-images.githubusercontent.com/985263/49203214-708fde00-f375-11e8-82cb-b7642bf3b4cf.png

Likewise for causatives: [image: image] https://user-images.githubusercontent.com/985263/49203298-be0c4b00-f375-11e8-9e40-37b002571de7.png

This is proving problematic for SNACS integration, because the overall event is a change-of-state, and thus should be a P. It also is problematic for paraphrasing: the aspectual type for "Mary got pregnant" (D, S) would differ from its paraphrase "Mary was impregnated" (P).

One solution that keeps these aspectual verbs as secondary Ds would be to annotate D and S within a P unit:

  • I_A [got_D angry_S]_P
  • John_A [made_D]_P- me_A [angryS]-P

Another would be to use multiple categories, e.g.:

  • I_A got_P+D angry_S
  • John_A made_P+D me_A angry_S

"P+D" could be interpreted as "introduces a change of state, but secondary to another predicate".

(Contrast resultatives like "hammered the metal flat", which have two primary scene-evokers, so both P and S are represented.)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/UniversalConceptualCognitiveAnnotation/docs/issues/50, or mute the thread https://github.com/notifications/unsubscribe-auth/AIG866BgcLB7wVAkpx_VB75MJ70aQAxOks5uz4EwgaJpZM4Y5CUq .