This frame has an issue in that it can have both a theme and a predicate: "Mary believed John that he didn't eat the last piece of pie". The problem is that we've been annotating both themes and predicates as arg1, so simply making either theme or predicate arg2 will conflict with past annotation. We could add an arg2, predicate, and use it only when arg1 is already present, but that makes me sad, so let's just not mess with it until we actually see such an instance. (from believe.01-v)
        Arg0-PAG: believer (vnrole: 29.9-2-Agent, 31.2-Experiencer, 29.5-1-Agent)
        Arg1-PPT: believed (vnrole: 29.9-2-Theme, 31.2-Stimulus, 29.5-1-Theme)
        Arg2-PRD: attribute, arg1 is believed to be what? (vnrole: 29.9-2-Attribute, 29.5-1-Predicate, 31.2-Attribute)
Example: nominal complement
        Cathryn Rice could hardly believe her eyes.
        Arg0: Cathryn Rice
        Argm-mod: could
        Argm-adv: hardly
        Rel: believe
        Arg1: her eyes
Example: prepositional complement
        I believe in the system.
        Arg0: I
        Rel: believe
        Arg1: in the system
Example: sentential complement
        You believe that Seymour Cray can do it again.
        Arg0: You
        Rel: believe
        Arg1: that Seymour Cray can do it again
Example: sentential complement with passive extraction
        [The declaration by Economy Minister Nestor Rapanelli]-1 is believed *-1 to be the first time such an action has been called for.
        Rel: believed
        Arg1: [The declaration by Economy Minister Nestor Rapanelli]-1
        Arg2: *-1 to be the first time such an action has been called for
Example: ARG0 and ARG1
        For that matter , the Chinese display a willingness to believe in the auspiciousness of just about anything !
        Arg0: the Chinese
        Rel: believe
        Arg1: in the auspiciousness of just about anything
Predicate: believable
Roleset id: believable.02 , able to be believed, Source: , vncls: , framnet:
believable.02: BELIEVABLE-J NOTES: Added by Julia based on BOLT-6, relation to believe.01, unbelievable.01. (from believable.01-j)
        Connected to this , last year India actually took some actions as to this , and these words should be believable . We are not wrong in analyzing it this way :
        Arg1: these words
        Argm-mod: should
        Rel: believable