search

UniversalDependencies / UD_English-GUM

Other
30 stars 4 forks source link

Secret Agent, Secret Service as an ADJ with a Number? #78

Open AngledLuffa opened 9 months ago

AngledLuffa commented 9 months ago

There are a couple times where Secret Service or Secret Agent have Secret tagged as UPOS ADJ and XPOS NNP. It has the features Degree=Pos|Number=Sing in that case. Perhaps that should just be Degree=Pos

# sent_id = GUM_interview_mcguire-6
# text = I launched LibriVox, emailed some friends and some podcasters who where doing literary stuff, and invited them to record a chapter of Joseph Conrad's Secret Agent.
28      Secret  Secret  ADJ     NNP     Degree=Pos|Number=Sing  29      amod    29:amod XML=<hi rend:::"italic">
29      Agent   Agent   PROPN   NNP     Number=Sing     23      nmod    23:nmod:of      Entity=27)26)|SpaceAfter=No|XML=</hi>

# sent_id = GUM_speech_maiden-12
# text = The presence of thousands of National Guardsmen, police, and Secret Service was a reminder that 2 weeks ago, in this very room and on those very steps ...
12      Secret  Secret  ADJ     NNP     Degree=Pos|Number=Sing  13      amod    13:amod Entity=(38-organization-new-cf9-1,2-sgl-United_States_Secret_Service
13      Service Service PROPN   NNP     Number=Sing     7       conj    4:nmod:of|7:conj:and    Entity=38)35)

related to https://github.com/UniversalDependencies/UD_English-PUD/issues/37

nschneid commented 9 months ago

Agreed, Number should be restricted to non-ADJ NNPs.

rhdunn commented 9 months ago

What about plural forms like "Attorneys General"? Are those "Attorneys/PROPN+NNPS General/ADJ+NNP"? -- Note the switched noun/adj positions. In which case the ADJ+NNP should never be plural.

nschneid commented 9 months ago

I think modifiers within a proper name that do not themselves have plural marking are NNP, not NNPS, even if the head is NNPS.

E.g., "United States" is "United/ADJ+NNP States/PROPN+NNPS" in EWT and "United/VERB+NNP States/PROPN+NNPS" in GUM.

amir-zeldes commented 9 months ago

E.g., "United States" is "United/ADJ+NNP States/PROPN+NNPS" in EWT and "United/VERB+NNP States/PROPN+NNPS" in GUM.

Agreed about the Number, but why is "United" ADJ in EWT? I thought this was consistent, aren't the States literally "United" with each other? If it means that they have undergone unionization, then shouldn't the lemma be "Unite"?

nschneid commented 9 months ago

It's the whole adjective/past participle difficulty. You can say a group is very united, for example.

amir-zeldes commented 9 months ago

Yes, see my comment just now here. ON has both tags for lowercased "united", though JJ is the majority tag. For the criteria in the link, the relative paraphrase decides it for VBN ("united people" can be "people who are united by ...")