Enhanced dependencies + ellipsis: Restoration of more than one verb

[bulbulistan]

It is my understanding that if clauses are coordinated and share the same elements, those can be omitted in the first and appear fully in the last such as:

I keep the girl away from the fire and you keep the cat away from the water.

In Middle Persian, a verb-final language, this would result in:

I the girl from the fire and you the cat from the water away keep.

UD only restores the verb here, which is counter-intuitive:

I the girl from the fire (keep) and you the cat from the water away keep.

instead of:

I the girl from the fire (away keep) and you the cat from the water away keep.

Are we allowed to restore more than the verb? Note that one or more restore elements would then depend on the restored root and they would not be "conj" to the existing root (which is the example given in the UD documentation).

Consider also the following example:

My father planned and built his house all by himself + my brother planned and built his house in the same way.

This would result in the Middle Persian structure:

My father his house all by himself planned and built and my brother in the same way.

Are we allowed to restore both verbs? I.e.:

My father his house all by himself planned and built and my brother in the same way (planned and built).

Can we indicate each verb by a consecutive number? This means: "planned" as .1 and "built" as .2 ?

[amir-zeldes]

Yes, I agree there should be multiple ellipsis nodes in cases like this, and I don't think there's any rule against it. UD_English-GUM has quite a few cases like that, for example:

1   Both    both    CCONJ   CC  _   2   cc:preconj  2:cc:preconj    Discourse=joint-other_m:45->43:2:_|Entity=(57-person-giv:inact-cf1-2,4-coref
2   Chao    Chao    PROPN   NNP Number=Sing 6   nsubj:pass  6:nsubj:pass    Entity=(6-person-giv:act-cf2*-1-coref-Yuen_Ren_Chao)
3   and and CCONJ   CC  _   4   cc  4:cc    _
4   Yang    Yang    PROPN   NNP Number=Sing 2   conj    2:conj:and|6:nsubj:pass Entity=(51-person-giv:inact-cf5-1-coref-Buwei_Yang_Chao)57)
5   were    be  AUX VBD Mood=Ind|Number=Plur|Person=3|Tense=Past|VerbForm=Fin   6   aux:pass    6:aux:pass  _
6   known   know    VERB    VBN Tense=Past|VerbForm=Part|Voice=Pass 0   root    0:root  MSeg=know-n
7   for for ADP IN  _   10  case    10:case _
8   their   their   PRON    PRP$    Case=Gen|Number=Plur|Person=3|Poss=Yes|PronType=Prs 10  nmod:poss   10:nmod:poss    Entity=(100-abstract-new-cf6-3-sgl(57-person-giv:act-cf1-1-ana)
9   good    good    ADJ JJ  Degree=Pos  10  amod    10:amod _
10  senses  sense   NOUN    NNS Number=Plur 6   obl 6:obl:for   MSeg=sense-s
11  of  of  ADP IN  _   12  case    12:case _
12  humor   humor   NOUN    NN  Number=Sing 10  nmod    10:nmod:of  Entity=100)|SpaceAfter=No
13  ,   ,   PUNCT   ,   _   14  punct   14:punct    _
14  he  he  PRON    PRP Case=Nom|Gender=Masc|Number=Sing|Person=3|PronType=Prs  6   parataxis   14.2:nsubj:pass Discourse=elaboration-additional:46->45:0:ref-prs-474,486+sem-lxchn-484,493,496|Entity=(6-person-giv:act-cf2*-1-ana-Yuen_Ren_Chao)
14.1    was be  AUX VBD Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin   _   _   14.2:aux:pass   CopyOf=-1
14.2    known   know    VERB    VBN Tense=Past|VerbForm=Part|Voice=Pass _   _   6:parataxis CopyOf=6
15  particularly    particularly    ADV RB  Degree=Pos  18  advmod  18:advmod   MSeg=particul-ar-ly
16  for for ADP IN  _   18  case    18:case _
17  his his PRON    PRP$    Case=Gen|Gender=Masc|Number=Sing|Person=3|Poss=Yes|PronType=Prs 18  nmod:poss   18:nmod:poss    Entity=(101-abstract-new-cf7-2-sgl(6-person-giv:act-cf2*-1-ana-Yuen_Ren_Chao)
18  love    love    NOUN    NN  Number=Sing 14  orphan  14.2:obl:for    _
19  of  of  ADP IN  _   21  case    21:case _
20  subtle  subtle  ADJ JJ  Degree=Pos  21  amod    21:amod Entity=(102-abstract-new-cf8-2-sgl
21  jokes   joke    NOUN    NNS Number=Plur 18  nmod    18:nmod:of  Entity=102)|MSeg=joke-s

[dan-zeman]

In the first example, I would also restore away although it is not directly the predicate. Maybe the Persian example should be added to the guidelines?

In the second example, I would restore both the verbs. Since they are predicates, this is even less questionable. I do not think we have been showing exactly this (coordination of verbs) but we definitely have examples where two verbs are restored because they are in xcomp relation (Susan wants to go to Prague and Jane [wants to go] to Rome.)

[bulbulistan]

Thank you both!