UniversalDependencies / UD_English-EWT

English data
Creative Commons Attribution Share Alike 4.0 International
199 stars 42 forks source link

"Every" instance that escaped? #533

Closed AngledLuffa closed 2 months ago

AngledLuffa commented 3 months ago

in the test set:

# sent_id = newsgroup-groups.google.com_JokeEruption_df151b356f94881c_ENG_20050819_155700-0027
# text = Every now and then she stops to breathe.
1       Every   every   ADV     RB      _       2       advmod  2:advmod        _
2       now     now     ADV     RB      PronType=Dem    6       advmod  6:advmod        _
3       and     and     CCONJ   CC      _       4       cc      4:cc    _
4       then    then    ADV     RB      PronType=Dem    2       conj    2:conj:and|6:advmod     _
5       she     she     PRON    PRP     Case=Nom|Gender=Fem|Number=Sing|Person=3|PronType=Prs   6       nsubj   6:nsubj|8:nsubj:xsubj   _
6       stops   stop    VERB    VBZ     Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin   0       root    0:root  _
7       to      to      PART    TO      _       8       mark    8:mark  _
8       breathe breathe VERB    VB      VerbForm=Inf    6       xcomp   6:xcomp SpaceAfter=No
9       .       .       PUNCT   .       _       6       punct   6:punct _

or was that intentional, such as it's an expression?

Also, there's this, which appears to be a typo:

# newdoc id = reviews-039173
# sent_id = reviews-039173-0001
# newpar id = reviews-039173-p0001
# text = The waiting staff is really friendly, its like every one knows each other, the manager is really sweet and the food..well no complaints from me.
11      every   every   DET     DT      PronType=Tot    12      det     12:det  _
12      one     one     NOUN    NN      Number=Sing     13      nsubj   13:nsubj        _
nschneid commented 3 months ago

Oh interesting, I think the issue is that "every" is normally a DET but in "every now and then" it is tagged ADV. #432

"every one": yeah maybe that should be goeswith despite it being a clear etymological origin of the pronoun. Would need to apply consistently to "some one", "any one" etc.

nschneid commented 3 months ago

EWT & GUM are consistently using det for some/any/every + one even though the spelling is a bit nonstandard. @amir-zeldes do you think it's worth changing to goeswith?

amir-zeldes commented 2 months ago

I think it depends on whether it's actually an error for "everyone" - the case I see in GUM isn't IMO, because it says "every one of your tribes" which I think can't be replaced by "everyone of your tribes" (maybe that's also grammatical, but would mean every person amongst those tribes, not "each tribe of those tribes"). If it's replaceable by "everyone" I'd be open to analyzing it as goeswith.

nschneid commented 2 months ago

GUM has "some one" and "any one" with det though—I think these are replaceable with the pronouns.

amir-zeldes commented 2 months ago

Ordinarily I'd say sure, let's make it goeswith, but I don't think these cases are replaceable with pronouns. They actually come from an official political speech transcript (so less likely to be typos), and I think it means something slightly different. Notice how both cases are partitive with "of", and can be replaced with "any single one of":

Regular "anyone" can't be replaced by "any single one":

I can't find the "some one" case, can you point me to it?

nschneid commented 2 months ago

Oh you're right. I was accidentally searching EWT yesterday when I found

That's in addition to:

I can replace those with goeswith.

amir-zeldes commented 2 months ago

Yes, those look right to me too