search

propbank / propbank-frames

Lexicon of frame files used by Propbank annotation. A searchable, readable version of the latest release is here: http://propbank.github.io/v3.4.0/frames/
Creative Commons Attribution Share Alike 4.0 International
95 stars 27 forks source link

roles for wonder.01 and wonder.02 #18

Open jheinecke opened 6 days ago

jheinecke commented 6 days ago

file wonder.xml contains the rolesets wonder.01 and wonder.02 With the roles

       <role descr="thinker" f="PPT" n="0">
          <rolelinks>
            <rolelink class="best_guess" resource="VerbNet" version="verbnet3.3">experiencer</rolelink>
            <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.3">experiencer</rolelink>
            <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.4">experiencer</rolelink>
          </rolelinks>
        </role>
        <role descr="thought" f="PAG" n="1">
          <rolelinks>
            <rolelink class="best_guess" resource="VerbNet" version="verbnet3.3">stimulus</rolelink>
            <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.3">stimulus</rolelink>
            <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.4">stimulus</rolelink>
          </rolelinks>
        </role>

I wonder whether the f-tag is correct ? Shouldn't it rather be

 <role descr="thinker" f="PAG" n="0">
 ...
 </role>
 <role descr="thought" f="PPT" n="1">
 ...
</role>

i.e. PAG as tag for argument 0. or at least not PAG as tag for argument 1

iccricket16 commented 4 days ago

Hi Johannes,

Thank you for your message. Upon review, this does not appear to be an error.

In wonder.01 (“John wondered what to do”), the Arg0 is an agent who consciously chooses to engage in the act of wondering. In wonder.02 (“John wondered at the display of unreasoning hatred”), the Arg0 is an experiencer who is “struck” by a sense of wonder. They aren’t choosing to experience the emotion. Instead, the Arg1 is the stimulus. In wonder.01, the Arg0 (agent) is the most agentive argument (PAG) and the Arg1 (theme) is the most patient-like argument (PPT) but in wonder.02 the Arg0 (experiencer) is the PPT and the Arg1 (stimulus) is the PAG. I will note that we do consider stimuli to be good candidates for PAG in general and more agentive than experiencers specifically. So, we would prioritize this function tag over something like CAU.

I hope this addresses your question,

Katie

On Jul 1, 2024, at 6:37 AM, Johannes Heinecke @.***> wrote:

 [External email - use caution]

file wonder.xml contains the rolesets wonder.01 and wonder.02 With the roles

   <role descr="thinker" f="PPT" n="0">
      <rolelinks>
        <rolelink class="best_guess" resource="VerbNet" version="verbnet3.3">experiencer</rolelink>
        <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.3">experiencer</rolelink>
        <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.4">experiencer</rolelink>
      </rolelinks>
    </role>
    <role descr="thought" f="PAG" n="1">
      <rolelinks>
        <rolelink class="best_guess" resource="VerbNet" version="verbnet3.3">stimulus</rolelink>
        <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.3">stimulus</rolelink>
        <rolelink class="care-88.1-1" resource="VerbNet" version="verbnet3.4">stimulus</rolelink>
      </rolelinks>
    </role>

I wonder whether the f-tag is correct ? Shouldn't it rather be

... ...

i.e. PAG as tag for argument 0. or at least not PAG as tag for argument 1

— Reply to this email directly, view it on GitHubhttps://github.com/propbank/propbank-frames/issues/18, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADKRQ5NUWX7I7IKWRDJ35XLZKFEPHAVCNFSM6AAAAABKFPWRJWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGM4DGNZTG44DQMI. You are receiving this because you are subscribed to this thread.Message ID: @.***>

jheinecke commented 4 days ago

Hi Katie, thanks for this info! I thought PAG is rather "actor" than just "most agentive". But, nevertheless, wouldn't be "CAU" better for the ARG1 ? Since the ARG1 can be rather unagentive (like in "John (ARG0/PPT) wondered how the wind broke the window"

iccricket16 commented 3 hours ago

 Hi Johannes,

Yes, our PAG stands for proto-agent and PPT stands for proto-patient. The most agentive-like argument is given the PAG function tag.

You bring up a good point about the Arg1. In this case, we’ve differentiated wonder.01 and wonder.02 in part on volitional engagement. Wonder.02 is used for cases in which it is clear that the Arg1 is a stimulus and Arg0 an experiencer. In such cases, the Arg0, while often sentient, does not have much control over the emotion they experience. In wonder.01, the Arg0 is certainly impacted by the Arg1 as the Arg1 directs the focus of their thoughts. However, it is assumed that the Arg0 has some degree of control over the choice to think about (or not) how the window was broken. So, in the case of wonder.01, the Arg0 is considered a PAG because they retain a degree of control (cause) over their cognitive engagement with the event.

Best, Katie

On Jul 3, 2024, at 8:40 AM, Johannes Heinecke @.***> wrote:

 [External email - use caution]

Hi Katie, thanks for this info! I thought PAG is rather "actor" than just "most agentive". But, nevertheless, wouldn't be "CAU" better for the ARG1 ? Since the ARG1 can be rather unagentive (like in "John (ARG0/PPT) wondered how the wind broke the window<ARG1/CAU>"

— Reply to this email directly, view it on GitHubhttps://github.com/propbank/propbank-frames/issues/18#issuecomment-2206344208, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADKRQ5M7NHPQFTXGQ2VXEDLZKQEMPAVCNFSM6AAAAABKFPWRJWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMBWGM2DIMRQHA. You are receiving this because you commented.Message ID: @.***>