UniversalDependencies / docs

Universal Dependencies online documentation
http://universaldependencies.org/
Apache License 2.0
272 stars 247 forks source link

obj vs. iobj vs. obl in Catalan, Spanish and Italian for the verbs "agradar", "gustar" and "piacere" #290

Open ftyers opened 8 years ago

ftyers commented 8 years ago

I've noticed what might be an inconsistency between how the verbs "agradar" (cat) and "gustar" (spa) are treated in Catalan and Spanish. In both languages they mean "like", and are used like "It likes to.me". In the Catalan treebank "to.me" gets the dobj label, while in the Spanish treebank, "to.me" gets the iobj label.

Other Romance languages: Portuguese has "gostar" which does not work the same way, and French has "aimer" which also does not. Italian has "piacere" which goes with iobj.

Spanish:

42      y       _       CONJ    _       _       35      cc      _       _
43      eso     _       PRON    _       Gender=Com|Number=Sing  45      nsubj   _       _
44      me      _       PRON    _       Gender=Com|Number=Sing|Person=1 45      iobj    _       _
45      gusta   _       VERB    _       Gender=Com|Number=Sing|Person=3|Mood=Ind|Tense=Pres     35      conj    _       _
46      .       _       PUNCT   _       _       1       punct   _       _

Catalan:

1       Li      ell     PRON    PRON    Case=Dat|Number=Sing|Person=3|PronType=Prs      2       dobj    _       _
2       agrada  agradar VERB    VERB    Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin   0       root    _       _
3       l'      el      DET     DET     Definite=Def|Number=Sing|PronType=Art   4       det     _       _
4       actualització   actualització   NOUN    NOUN    Gender=Fem|Number=Sing  2       nsubj   _       _
1       A       a       ADP     ADP     AdpType=Prep    4       case    _       _
2       tots    tot     DET     DET     Gender=Masc|Number=Plur|PronType=Ind    3       det     _       _
3       els     el      DET     DET     Definite=Def|Gender=Masc|Number=Plur|PronType=Art       4       det     _       _
4       actors  actor   NOUN    NOUN    Gender=Masc|Number=Plur 6       dobj    _       _
5       els     ell     PRON    PRON    Number=Plur|Person=3|PronType=Prs       6       dobj    _       _
6       agrada  agradar VERB    VERB    Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin   0       root    _       _
7       conèixer        conèixer        VERB    VERB    VerbForm=Inf    6       csubj   _       _
2       L'      el      DET     DET     Definite=Def|Number=Sing|PronType=Art   6       det     _       _
3       única   únic    ADJ     ADJ     Gender=Fem|Number=Sing  6       amod    _       _
4       que     que     PRON    PRON    PronType=Rel    6       dobj    _       _
5       m'      em      PRON    PRON    Number=Sing|Person=1|PronType=Prs       6       dobj    _       _
6       agrada  agradar VERB    VERB    Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin   10      csubj   _       _
7       veure   veure   VERB    VERB    VerbForm=Inf    6       csubj   _       _

Italian:

1       "       "       PUNCT   FB      _       4       punct   _       _
2       Non     non     ADV     BN      PronType=Neg    4       neg     _       _
3       mi      mi      PRON    PC      Number=Sing|Person=1|PronType=Clit      4       iobj    _       _
4       piace   piacere VERB    V       Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin   0       root    _       _
5       il      il      DET     RD      Definite=Def|Gender=Masc|Number=Sing|PronType=Art       12      det     _       _
6       suo     suo     DET     AP      Gender=Masc|Number=Sing|Poss=Yes|PronType=Prs   12      det:poss        _       _
7       voler   volere  AUX     VM      VerbForm=Inf    12      aux     _       _
8       essere  essere  VERB    V       VerbForm=Inf    12      cop     _       _
jipatsaa commented 8 years ago

I think you are right, according to the DIEC (Diccionari De l'Institut
d'Estudis Catalans) or (Fábregas etal., 2012: 164), there are three
kind of psychological verbs depending on the syntactic position
occupied by the experiencer:

Francis Tyers notifications@github.com erabiltzaileak idatzi du:

I've noticed what might be an inconsistency between how the verbs ...

ftyers commented 8 years ago

I think that from the traditional grammatical sense, these are definitely iobj, however from a UD case, it isn't so clear. In terms of these verbs, they cannot have an "accusative" object, so, the experiencer here could be considered the "second most core argument after the subject" (as in the definition of dobj below).

It would be good to hear from the people making the treebanks, incidentally in Basque the experiencer of 'gustatu' gets iobj.)

dobj:

The direct object of a verb is the second most core argument of a verb after the subject. Typically, it is the noun phrase that denotes the entity acted upon or which undergoes a change of state or motion (the proto-patient).

iobj:

The indirect object of a verb is any nominal phrase that is a core argument of the verb but is not its subject or direct object. The prototypical example is the recipient of ditransitive verbs of exchange:

hectormartinez commented 8 years ago

I prefer the dobj analysis, regardless of the experiencer being marked in dative.

jipatsaa commented 8 years ago

In the Basque treebank, and following several authors (Oyarzabal 2003,
Artiagoitia 1995, 2000) who analyzed Basque psychological verbs in the
past, we tagged the experiencer as iobj as next sentence shows.

1 Bikiei biki NOUN Case=Dat|Definite=Def|Number=Plur 3 iobj 2 gehien gehien ADV 3 advmod 3 gustatzen gustatu VERB Aspect=Imp|VerbForm=Inf 5 acl 4 zaien izan AUX Mood=Ind|Number[abs]=Sing|Number[dat]=Plur|Person[abs]=3|Person[dat]=3 3 aux 5 lekua leku NOUN Animacy=Inan|Case=Abs|Definite=Def|Number=Sing 7 nsubj 6 sukaldeko sukalde NOUN Animacy=Inan|Case=Loc|Definite=Def|Number=Sing 7 nmod 7 balkoia balkoi NOUN Animacy=Inan|Case=Abs|Definite=Def|Number=Sing 0 root 8 da izan VERB Aspect=Prog|Mood=Ind|Number[abs]=Sing|Person[abs]=3 7 cop 9 , , PUNCT 3 punct _

I hope it helps to clarify the reason for that decission.

Here one reference (B. Oyharçabal. Lexical causatives and causative
alternation in Basque. XLVI, Universitad del pais Vasco - Euskal
Herriko Unibertsitatea, pp.223-253, 2003):

Adhering to the typology usually applied to these verbs (Belleni &
Rizzi 1988),ahaztu and gustatu belong to the piacere type of
psych-verb (Artiagoitia 1995, 2000). Such verbs do not admit a lexical causative alternation, as the
following examples show: (9a) Adinarekin kantuak ahaztu zaizkit age.COM song.PL.ABS forget AUX:3PL.1SG "On account of age I have forgotten the songs." (9b) * Adinak kantuak ahaztu dizkit age.ERG song.PL.ABS forget AUX:3SG.3PL.l SG *"Age has forgotten}TIe the songs." (9b ') Adinak kantuak ahatzarazi dizkit age.ERG song.PL.ABS forget.CAU AUX:3SG.3PL.1SG "Age has made me forget the songs.

jipatsaa commented 8 years ago

I guess that the definition

dobj:

The direct object of a verb is the second most core argument of a verb after the subject. Typically, it is the noun phrase that denotes the entity acted upon or which undergoes a change of state or motion (the proto-patient).

comes from the thematic role hierarchy defended among others by Dowty
(1991) as a way to capture generalities. But psychological verbs are
complex in lexico-semantic terms, but in the romance treebanks you
mentioned they used "iobj", we also used it in Basque so I would say
that the description appearing in the UD "dobj" tag means to capture a

generalization that might not apply to certain psy verbs.

dan-zeman commented 11 months ago

This issue was started under UD v1 guidelines, so dobj is now obj, but some of the inconsistencies still persist today (UD 2.13).

For agradar in Catalan, see this query: http://hdl.handle.net/11346/PMLTQ-NOAI. Pronominal experiencers are still often attached as obj because their form is ambiguous between dative and accusative. However, full nominals use the preposition a and they are obl:arg.

For gustar in Spanish AnCora, see this query: http://hdl.handle.net/11346/PMLTQ-R9HO. Pronominal experiencers are attached as obj even when they are unambiguously dative (le, les). They should probably be obl:arg because dative is not a core case in Spanish. Interestingly, me gusta is obl:arg although me could be dative or accusative. In Spanish GSD (query http://hdl.handle.net/11346/PMLTQ-RHM1), the experiencers are correctly obl:arg.

I suppose the experiencers of Italian piacere would also be best (most UD-way) analyzed as obl(:arg). I haven't checked all the Italian treebanks but the query for ISDT is http://hdl.handle.net/11346/PMLTQ-FOXO. Experiencers with preposition are obl here, bare pronouns are iobj (the pronouns are not annotated with the Case feature).

dan-zeman commented 5 months ago

UD 2.14:

No change in Catalan (new query: http://hdl.handle.net/11346/PMLTQ-AZYU).

In Spanish AnCora (new query: http://hdl.handle.net/11346/PMLTQ-BHEB), the deprels of dative pronouns have been fixed to obl:arg.