validator - event, argument structure and aspect

[Marketa-Lopatkova]

1. Predicative nouns have their own argument structure but not aspect
   Czech example: podnět (as in podal podnět vůči něčemu 'to file a complaint against something/somebody') can be seen as a predicative noun that form the meaning of the whole light verb construction. It is the noun which brings its participant to the structure, so it it should have its own argument structure. At the same time, it is not clear which lexical verb can be used instead of the light verb construction.
   On the other hand, predicative nouns does not typically encode aspect.
   English example: note (as a noun) is listed in the PropBank with ARG1 (frame note-02), with 2 examples note
   of the growing problem of computer fraud and little note of this development
   https://verbs.colorado.edu/propbank-development/note.html#note.02

2. Abstract predicate say-91
   Thus abstract predicate should "be used for unattributed speech in dialogues, etc-- anywhere there is not a predicate of speaking in the surface sentence. Also for 'according to X, Y', in English" (from UMR lists).
   This, it is probably the right way how to annotate prý 'reportedly' in Czech.
   However, in this case, it has no aspect value?

3. event as a place-holder
   The PDT2UMR automatic conversion uses the "event" label to replace #EmpVerb. The aspect attribute seems not relevant here (the same for modal-strength).

[dan-zeman]

1. Predicative nouns have their own argument structure but not aspect
   Czech example: podnět (as in podal podnět vůči něčemu 'to file a complaint against something/somebody') can be seen as a predicative noun that form the meaning of the whole light verb construction. It is the noun which brings its participant to the structure, so it it should have its own argument structure. At the same time, it is not clear which lexical verb can be used instead of the light verb construction.
   On the other hand, predicative nouns does not typically encode aspect.
   English example: note (as a noun) is listed in the PropBank with ARG1 (frame note-02), with 2 examples note
   of the growing problem of computer fraud and little note of this development
   https://verbs.colorado.edu/propbank-development/note.html#note.02

So what is the suggestion for the validator?

2. Abstract predicate say-91
   Thus abstract predicate should "be used for unattributed speech in dialogues, etc-- anywhere there is not a predicate of speaking in the surface sentence. Also for 'according to X, Y', in English" (from UMR lists).
   This, it is probably the right way how to annotate prý 'reportedly' in Czech.
   However, in this case, it has no aspect value?

Why not?

3. event as a place-holder
   The PDT2UMR automatic conversion uses the "event" label to replace #EmpVerb. The aspect attribute seems not relevant here (the same for modal-strength).

Is it irrelevant or just unknown to the automatic conversion procedure?

[Marketa-Lopatkova]

1. Predicative nouns have their own argument structure but not aspect
   Czech example: podnět (as in podal podnět vůči něčemu 'to file a complaint against something/somebody') can be seen as a predicative noun that form the meaning of the whole light verb construction. It is the noun which brings its participant to the structure, so it it should have its own argument structure. At the same time, it is not clear which lexical verb can be used instead of the light verb construction.
   On the other hand, predicative nouns does not typically encode aspect.
   English example: note (as a noun) is listed in the PropBank with ARG1 (frame note-02), with 2 examples note
   of the growing problem of computer fraud and little note of this development
   https://verbs.colorado.edu/propbank-development/note.html#note.02

So what is the suggestion for the validator?

The validator should not expect that whatever has an argument structure as an event (and thus has to have the "aspect" attribute). Now it complains that predicative nouns do not have aspect annotation.

2. Abstract predicate say-91
   Thus abstract predicate should "be used for unattributed speech in dialogues, etc-- anywhere there is no predicate of speaking in the surface sentence. Also for 'according to X, Y', in English" (from UMR lists).
   This, it is probably the right way how to annotate prý 'reportedly' in Czech.
   However, in this case, it has no aspect value?

Why not?

Aspect is irrelevant if say-91 refers to a particle, I believe.
(If yes - which one?)

3. event as a place-holder
   The PDT2UMR automatic conversion uses the "event" label to replace #EmpVerb. The aspect attribute seems not relevant here (the same for modal-strength).

Is it irrelevant or just unknown to the automatic conversion procedure?

Probably irrelevant - #EmpVerb is used in a verbal clause (as the head of the clause) in cases when a governing verb cannot be recovered from the context.

[dan-zeman]

The validator should not expect that whatever has an argument structure as an event (and thus has to have the "aspect" attribute). Now it complains that predicative nouns do not have aspect annotation.

According to the guidelines, they should have aspect. The value should be process.

[dan-zeman]

Aspect is irrelevant if say-91 refers to a particle, I believe.
(If yes - which one?)

I think that process is relevant here as well. If not even perfective (it always means that someone has said something, doesn't it?)

[dan-zeman]

5. event as a place-holder
The PDT2UMR automatic conversion uses the "event" label to replace #EmpVerb. The aspect attribute seems not relevant here (the same for modal-strength).

Is it irrelevant or just unknown to the automatic conversion procedure?

Probably irrelevant - #EmpVerb is used in a verbal clause (as the head of the clause) in cases when a governing verb cannot be recovered from the context.

Regarding this item, it is difficult to base a decision on the guidelines, since the very existence of event is our invention that goes beyond the guidelines. But human annotators are not supposed to use any placeholders. They have to imagine an event, so they should imagine it together with its aspect, even if blurred, shouldn't they?

[Marketa-Lopatkova]

5. event as a place-holder
   The PDT2UMR automatic conversion uses the "event" label to replace #EmpVerb. The aspect attribute seems not relevant here (the same for modal-strength).

Is it irrelevant or just unknown to the automatic conversion procedure?

Probably irrelevant - #EmpVerb is used in a verbal clause (as the head of the clause) in cases when a governing verb cannot be recovered from the context.

Regarding this item, it is difficult to base a decision on the guidelines, since the very existence of event is our invention that goes beyond the guidelines. But human annotators are not supposed to use any placeholders. They have to imagine an event, so they should imagine it together with its aspect, even if blurred, shouldn't they?

The abstract concept event is listed (as one of the basic concepts) in the UMR lists, so it is not just our invention (the lists are kept updated, contrary to the guidelines). A bit surprisingly, it should have refer menu as other entities (person, thing, place).
In the UMR 1.0 English data, the event concept is used, e.g., as :ARG1 of the event survive-02 (for ... spattered survivors were also shown, english_umr-0001.txt, snt 8); neither survive, nor event has aspect here (and the event concept is not marked as coreferential, so it does not inherit the aspect value from its antecedent). Of course, the example might be incomplete - but all occurrences of the event concept are left WITHOUT aspect in the English 1.0 data (but also WITHOUT refer-.*), in several cases, it has :mod relation)!
Another interesting example (english_umr-0004.txt, snt34): And you watch the goat , ... is annotated as you watch [an event that is underwent by] the goat ,

[Marketa-Lopatkova]

# sentence level graph:
(s34w / watch-01
  :ARG0 (s34p / person
          :refer-person 2nd
          :refer-number singular)
  :ARG1 (s34e / event
          :undergoer (s34g / goat
                  :refer-number singular))
  :aspect state)