Is this "yours" also genitive case? (Maybe none of them?)

[AngledLuffa]

Two examples of yours are genitive:

# sent_id = GUM_vlog_exams-136
# text = The floor is yours.
4       yours   yours   PRON    PRP$    Case=Gen|Number=Sing|Person=2|Poss=Yes|PronType=Prs     0       root    0:root  Entity=(106-person-giv:inact-cf1-1-ana)|SpaceAfter=No

# sent_id = GUM_vlog_pizzeria-62
# text = You want onions on yours?
5       yours   yours   PRON    PRP$    Case=Gen|Number=Sing|Person=2|Poss=Yes|PronType=Prs     2       obl     2:obl:on        Entity=(25-person-giv:act-cf1-1-ana)|SpaceAfter=No

This one is different:

# sent_id = GUM_whow_languages-6
# s_prominence = 3
# s_type = decl
# transition = continue
# text = The choice is all yours.
1       The     the     DET     DT      Definite=Def|PronType=Art       2       det     2:det   Discourse=restatement-partial:8->5:2:ref-prs-22,57|Entity=(12-event-acc:com-cf2-2-sgl
2       choice  choice  NOUN    NN      Number=Sing     5       nsubj   5:nsubj Entity=12)
3       is      be      AUX     VBZ     Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin   5       cop     5:cop   _
4       all     all     ADV     RB      Degree=Pos      5       advmod  5:advmod        _
5       yours   your    PRON    PRP     Number=Sing|Person=2|Poss=Yes|PronType=Prs      0       root    0:root  Entity=(5-person-giv:act-cf1*-1-ana)|SpaceAfter=No
6       .       .       PUNCT   .       _       5       punct   5:punct _

[AngledLuffa]

A past PR on the Pronouns dataset had removed its genitive cases:

https://github.com/UniversalDependencies/UD_English-Pronouns/pull/5

but there are quite a few Case=Gen in EWT

[AngledLuffa]

Also, the lemmas of yours don't agree with the current PRON table:

https://universaldependencies.org/en/pos/PRON.html

although further inspection of that table makes me think none of these instances should be Case=Gen, rather than all of them

[amir-zeldes]

The root cause is that this is an xpos tag error - PRP$ should be PRP. Will fix, thanks!

[AngledLuffa]

Didn't even look at that, nice catch. Still, might be worth rechecking the features as well

[amir-zeldes]

The features are auto generated based on the manual annotations, and in this case the culprit is the xpos tag (and the lemma would have tripped the validator if xpos had been correct)

[AngledLuffa]

What I mean is maybe none of them should be Case=Gen according to the PRON page

https://universaldependencies.org/en/pos/PRON.html

[amir-zeldes]

Indeed, as long as they are not xpos=PRP$, they will also not be Case=Gen:

https://github.com/amir-zeldes/gum/blob/master/_build/utils/ud_morph.ini#L59

[AngledLuffa]

The first couple examples in this thread were Case=Gen, but perhaps you mean it will change now

[amir-zeldes]

Yup, that will change automatically. Almost all FEATS in GUM are refreshed and generated automatically with each release, so there is no need to correct them manually.

[rueter]

# sent_id = GUM_vlog_pizzeria-62
# text = You want onions on yours?
5       yours   yours   PRON    PRP$    Case=Gen|Number=Sing|Person=2|Poss=Yes|PronType=Prs     2       obl     2:obl:on        Entity=(25-person-giv:act-cf1-1-ana)|SpaceAfter=No

Here yours should not be labled Case=Gen, unless the adposition takes genitive goverment. Lemmatization looks correct if we consider function. @amir-zeldes

[amir-zeldes]

Here yours should not be labled Case=Gen

Indeed, and now that the underying xpos has been changed to PRP it will not happen - as soon as FEATS are regenerated, it will not be Case=Gen anymore.

Lemmatization looks correct if we consider function

No, it was decided to split these and the personal pronouns into exactly two lemmas - the attributive possessive form for all the possessives (your, yours -> your) and the nominative personal pronoun for the rest (you), so the correct lemma is "your" (as now corrected upstream). See the whole specification here:

https://universaldependencies.org/en/pos/PRON.html