deva3105
Joined: 11 May 2011
Posts: 4
|
 Posted: Fri Jul 01, 2011 6:25 am Post subject: UNL inadequacy? |
 |
|
Hey,
| Quote: |
[S:1239]
{org:fr}
Les enfants seuls savent ce qu'ils cherchent, fit le petit prince.
{/org}
{unl}
aoj(200595630:11, 109917593:12.@only.@pl)
obj(200595630:11, 00:15.@wh)
agt(201315613:78, 109917593:12.@only.@pl)
obj(201315613:78, 00:15.@wh)
agt(201755816:28.@past.@entry, Petit Prince:79)
obj(201755816:28.@past.@entry, 200595630:11)
{/unl}
[/S] |
The above UNL sentence is from the corpus of "The little Prince". The intended meaning is "Only the children know what they are looking for, said the little prince." But the UNL doesn't seem to be able to differentiate between the intended sentence and the sentence - "Only the children know whom they are looking for, said the little prince." The difference lies in the animate quality of the object and since the object is not actually known in a question, is the question not underspecified?
Or am I missing something here?
On a side note, the English grammar doesn't seem to have a rule for the last relation in the above UNL sentence (when relation "obj" acts on two verbs)
Thanks
Deva
Invention Labs |
|
martins
Site Admin
Joined: 16 Dec 2009
Posts: 1481
Location: Geneva, Switzerland
|
| Posted: Mon Jul 04, 2011 5:39 pm Post subject: RES: UNL inadequacy? |
|
|
Dear Deva,
In the UNL representation for « Les enfants seuls savent ce qu’ils cherchent, fit le petit prince », the interrogative pronoun “ce que” is represented by the null UW (00) followed by the wh marker (@wh). Actually, the UW 00.@wh corresponds to all English interrogative pronouns: “who”, “whom”, “what”, “where”, “when” and “how”. The actual role is defined by the relation: “plc”, in case of “where”; “tim”, in case of “when”; “man”, in case of “how”; “agt”, in case of “what” and “who”; and “obj”, in case of “what” and “whom”. In that sense, I don’t see any problem in the representation: 00.@wh , as an argument of “obj”, can be either “what” and “whom”, which seems to be the case of the French sentence, where it’s not possible to state that the children search for an animate or an inanimate being.
If one was sure that the children were searching for someone, instead of something, the representation should be either 00.@person.@wh or 00.@thing.@wh , but, again, this seems not to be the case here, where the source sentence is ambiguous.
As for the question concerning the English grammar, I’m not sure I have fully understood the issue that you have raised below. You mention that there is no rule, in the grammar, for “obj” acting on two verbs. I couldn’t understand why this would be necessary. Did you mean that the rule that applies over one relation is not valid for the other (even when they play exactly the same syntactic role, i.e., of a complement of a verb)? Or did you mean that there is no rule for verbs with two objects (i.e., ditransitive verbs, but this seems not be the case either)? Notice, please, that 1) the English grammar that can be exported from the UNLarium is still under construction (we depend on the stable releases of EUGENE and IAN, our analysis and generation tools, in order to tune it); and 2) the grammar there doesn’t cover relative clauses yet.
Best,
---------------------------------------------
Ronaldo MARTINS
Language Resources Manager
UNDL Foundation
48, route de Chancy
CH-1213 – Geneva - Switzerland
+41 22 879 8090
http://www.undlfoundation.org
---------------------------------------------
|
|
FAQ 
Search 
Memberlist 
Usergroups 
Profile 
