carmls / snacs-guidelines

Semantic Network of Adposition and Case Supersenses: Annotation Guidelines
2 stars 0 forks source link

X IN_GENERAL #25

Open nschneid opened 4 years ago

nschneid commented 4 years ago

Pets IN general are a problem for my allergies.

Is this pragmatic/metalinguistic?

[Reddit task_000:1080]

nschneid commented 4 years ago

Is it like inverse-Species?? cf. some kinds OF pets

Also, what about PER SE? "It's not pets PER_SE that are the problem"—this definitely feels metalinguistic (i.e. "pets" is the wrong way to describe the problem).

nschneid commented 4 years ago

cf. STREUSLE "a business OF this sort", which is inverse Species

nschneid commented 4 years ago

Could be serving a couple of functions:

svivek commented 4 years ago

Pets IN particular are a problem for my allergies.

Related to the second bullet above.

nschneid commented 4 years ago

Maybe better stated as: quantification that may have additional pragmatics attached.

svivek commented 4 years ago

Are these paraphrases?

svivek commented 4 years ago
nschneid commented 4 years ago

Hypothesis:

jenahwang commented 3 years ago

It sort of fits the definition for which Focus label was created. The adposition evokes the existence of a larger (unmentioned) set of things that the item belongs to, out of which the item is deemed most or least applicable to the event/situation described.

Looks like I said something similar for "At_least" #19.

esmanning commented 3 years ago

By Focus label do you mean `d? Or is there a secret extra supersense I don't know about?

jenahwang commented 3 years ago

There's a new label called "Focus" under Context tree that doesn't officially exist in Xposition. As to when we will make it official... maybe we should talk about this during the next meeting! It seems there are plenty uses for Focus in English as well.

On Thu, Aug 27, 2020 at 11:59 AM Emma Manning notifications@github.com wrote:

By Focus label do you mean `d? Or is there a secret extra supersense I don't know about?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/carmls/snacs-guidelines/issues/25#issuecomment-682133342, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANJDBVFBZEFJ6IDXBW6OQCTSC2UHLANCNFSM4IMOAOUQ .

-- Jena D. Hwang

Research Engineer, AI2

esmanning commented 3 years ago

Oh, cool! For the current reddit corpus release we still need a decision on how to annotate 'people IN general' that's compliant with guidelines v2.5, though.

nschneid commented 3 years ago

`d seems like a reasonable compromise for v2.5 guidelines.

nschneid commented 3 years ago

RLG thinks it's a hedge: https://twitter.com/rleegoldman/status/1373187773335674883

AH says quantifier: https://twitter.com/ah__cl/status/1373257871853559811