-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
"the judge": scene-evoking? #25
Comments
We discussed this a few days ago with one of the annotators: we said the profession should be annotated just P (without A or IMP A). |
So what are the criteria for annotating P+A or S+A on a unit? FrameNet has a concept of "denoted frame element"—basically if a noun both evokes a scene and denotes one of its roles. I figured that's what P/S+A was for. |
Resolution: judge should be P+A, not if it's a title. Cite FrameNet in a footnote. |
What about "the handwritten letter to John"? Is "letter" scene-evoking because it implies sending a message to John? |
One option: the_E [handwritten_P (letter)_A]_E letter_C [(IMP)_P [to_R John_C]_A]_E |
One of the annotators asked me: I saw the latter being used on several occasions also by annotators, This will make it similar to how we generally tend to treat these types of determiners in noun scenes : |
I agree with you. The first option is better.
…On Wed, Oct 3, 2018 at 10:15 AM dotdv ***@***.***> wrote:
One of the annotators asked me:
[[[the_E judge_C]_P+A]_A listened to the witness.
or
[[the]_E [judge_P+A]_C]_A listened to the witness.
I saw the latter being used on several occasions also by annotators,
but I think that the first option is more consistent with other
instructions we give, namely that Cs can't be scenes and that determiners
should be included inside the P/S (articles and demonstratives only I
think).
So in a more complex example:
[[The_E]*{P+A-} tired_D [judge_C]*{-P+A}]_A listened to the witness
This will make it similar to how we generally tend to treat determiners in
noun scenes :
We saw_P [[a_E]*{P-} great_D [show_C]*{-P}]_A
This decision of still including the determiner within the P/S even when
there's an adjective (D in this case) in between is something we verbally
agreed upon with the Eng annotators, but haven't yet clarified in the
guidelines.
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#25 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIG86yzkYfiXffl7PvhqdB6vSUzbV59Tks5uhGQAgaJpZM4WoF_G>
.
|
There is a problem though. How should we deal with cases like: Maybe a possible alternative is to internally mark the Remote as P: "the_E [tall_S (judge_P)_A ]_E [judge]_C" |
Good point. Let's make it even more interesting and add a modifier that should go in the scene evoked by "judge". Suppose we're talking about a dog show, and thus refer to a "dog judge" (somebody who judges the dogs in the competition).
|
I think we should allow C Scenes, but instruct annotators to use them very
sparingly. For instance,
experienced_D taxi_A driver_P+A
and *not*:
experienced_E [taxi driver]_C
…On Thu, Oct 4, 2018 at 5:48 PM Nathan Schneider ***@***.***> wrote:
Good point. Let's make it even more interesting and add a modifier that
should go in the scene evoked by "judge".
Suppose we're talking about a dog show, and thus refer to a "dog judge"
(somebody who judges the dogs in the competition).
-
[[the_E [tall_S (judge)_A]_E judge_C]_P dog_A (judge)_A]_A was able to
reach...
Is this too complex of a P?
-
[the_E [tall_S (judge)_A]_E [dog_A judge_P+A]_C]_A was able to reach...
This feels like the more natural structure given the compositionality
of the sentence. But it has a C-scene.
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#25 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIG86xiNOr5PcOoeWXhXTjxrCCLNosebks5uhh_NgaJpZM4WoF_G>
.
|
OK, so
But
? |
Yes. Exactly.
…On Thu, Oct 4, 2018 at 6:14 PM Nathan Schneider ***@***.***> wrote:
OK, so
- the driver: [the_E driver_C]_P+A
- the taxi driver: taxi_A [the_E driver_C]_P+A
- the driver of the taxi: [the_E driver_C]_P+A [of_R the_E taxi_C]_A
- the experienced taxi driver: experienced_D taxi_A [the_E
driver_C]_P+A (presumably "experienced" relates to the manner of driving)
But
- the tall taxi driver: the_E [tall_S (driver)_A]_E [taxi_A
driver_P+A]_C
- the taxi driver who swims: the_E [taxi_A driver_P+A]_C [who_R
[swims_P (driver)_A]_C]_E
?
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#25 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIG86zcUWTGO45GuV1g-C8jLb8PKIlTQks5uhiW8gaJpZM4WoF_G>
.
|
added the option of "C-scenes" under subsection "Model 3: Inter-Scene relations" |
A consequence of this is we don't really know whether the participant-entity or the scene itself is the head of the unit (implicitly, a normal Scene unit is headed by a P/S). Thus "the driver scared her" and "the driving scared her" would be structurally similar:
Are we OK with that? I guess the assumption could be that in a P+A unit, the A is the category that applies to the full word: "driver" as a whole refers to the person. Presumably zero-derived role words like "boss" and "engineer" would have P+A for the noun (entity) use and just P for the verb (event) use. What about more complicated derivational morphology where a noun for a phenomenon or quality is derived from a participant: "consumerism", "criminality", "demagoguery", "musicianship"? Should these be simply P/S, ignoring the fact that they are derived because the participants are generic? |
So far we know that a role/profession should first be marked a C-scene if one of its siblings does not modify the process that the role evokes (tall judge or judge who swims). But is it also true to say that if a role/profession is the head of a RC we necessarily have to mark the role C? And the next question would be what to do if the same scene appears as a top level scene? |
I agree with both analyses you suggested.
…On Mon, Nov 19, 2018 at 9:58 AM dotdv ***@***.***> wrote:
added the option of "C-scenes" under subsection "Model 3: Inter-Scene
relations"
*mentioned that they should be used sparingly
So far we know that a role/profession should first be marked a C-scene if
one of its siblings does not modify the process that the role evokes (tall
judge or judge who swims).
But is it also true to say that if a role/profession is the head of a RC
we necessarily have to mark the role C?
Related streusle example: Davie_A is_S [[a_F [patient_C and_N
methodical_C]_D teacher_P+A]_C [who has a great ear and sensitivity for his
students passions]_E]_A?
And the next question would be what to do if the same scene appears as a
top level scene?
[patient_C and_N methodical_C]*D teacher_P+A [who has a great ear...]*
?]_H?
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#25 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIG861yrbq-IbxHivgBSOOHYfIt3OX50ks5uwmSxgaJpZM4WoF_G>
.
|
Looks OK to me, though I worry we could get cases like "tall and patient teacher" where only one adjective describes the act of teaching. I don't understand how the second example is different. |
A student-authored example that we felt fairly confident about, except for one issue: p. 18 of the guidelines mentions that "driver" is a scene-evoking noun in "taxi driver". So should all uses of "judge" (the profession) be scene-evoking? Should it be P+A, because "judge" denotes both the scene and one of its participants? Should there be an implicit second participant, that which is judged?
The text was updated successfully, but these errors were encountered: