Copular constructions

[omriabnd]

We decided that in copula constructions the distinction should not be between copula objects that define a single identity and those that define a set, but rather between objects that define a Scene and ones that don't. So a "car" is not scene-evoking, but an "author" or "father" are. The subject is an A, and so the scene-evoking noun is just an S (not S+A).

Examples in ensuing photo.

[nschneid]

[nschneid]

[[a_E father_C]_S+A]_A arrived_P [my_A father_S+A]_A arrived_P vs. John_A is_F [a_E father_C]_S John_A is_F her_A father_S

[nschneid]

Another fun thing to figure out: when a copula relates two scenes, or a scene and a pronoun whose referent is a scene. Should the copula evoke an S in that case? Or should it be considered an L?

• That is a concert.
  a) That_A is_S [[a_C concert_E]_P]_A b) That_H is_L [[a_C concert_E]_P]_H

• That concert is a rock concert. a) [That_E concert_P]_A is_S [[a_C concert_E]_P rock_A]_A b) [That_E concert_P]_H is_L [[a_C concert_E]_P rock_A]_H

• Laughing is an attempt to avoid crying. a) [Laughing_P]_A is_S [[an_E attempt_C]_D to_F avoid_D crying_P]_A b) [Laughing_P]_H is_L [[an_E attempt_C]_D to_F avoid_D crying_P]_H

The (a) options feel more natural to me. The policy for "X copula Y" could then be: if Y is scene-evoking AND takes X as a participant, then the copula is F; otherwise, the copula is S, with X and Y as participants.

[nschneid]

And we need a policy for relative clauses:

• the fox, which is a mammal the_E fox_C [which_R (fox)_A is_S [a_E mammal_C]_A]_E

• John, who is her father John_C [who_R (John)_A is_F her_A father_S]_E

• That concert, which is a rock concert That_E concert_C [which_R (concert)_A is_S [[a_E concert_C]_P rock_A]_A]_E

  I don't know how to express the relative clause if the copula is a linker.

[dotdv]

We decided that in copula constructions the distinction should not be between copula objects that define a single identity and those that define a set, but rather between objects that define a Scene and ones that don't. So a "car" is not scene-evoking, but an "author" or "father" are. The subject is an A, and so the scene-evoking noun is just an S (not S+A). Examples in ensuing photo.

OK. I have only one question about the second whiteboard example: [This]_A is_S [a_E [red_S (car)_A]_E car_C]_A

The same example appears in the guidelines with an instruction to give precedence to the scene-evoking relation: P13: "If both alienable and inalienable attributions appear in the same scene the alienable attribution will take precedence and we will mark it as the S: ''This_A is_F [anE]{A-} amazing_S [restaurantC]{-A}'' ''This_A is_F [aE]{A-} red_S [carC]{-A}''

[nschneid]

What is the basis for not treating alienable attributions in the predicate complement as elaborators?

[dotdv]

When we tried to deal with such examples back then we had indeed already established that when there are no scene evoking elements like in "this is a car" the copula should be the S. But a case like "this is an amazing car", when seen as a whole, seemed essentially much more similar to "this car is amazing" where it is easy to fill in the P/S spot with "amazing" as the S.

[nschneid]

But a case like "this is an amazing car", when seen as a whole, seemed essentially much more similar to "this car is amazing" where it is easy to fill in the P/S spot with "amazing" as the S.

Ah, OK—I think there are two problems with this approach:

The practical problem is that it becomes awkward if there's more than one adjective. Consider "This is a nice red car and that is an ugly blue car." Under this policy I suppose you'd have 4 parallel scenes, whereas the syntax suggests 2 parallel scenes.

The theoretical problem is that "This is an amazing car" and "This car is amazing" may be interchangeable, but only in certain contexts with certain information structure. Consider "This is a black cat and that is a black dog", where it is the nouns that contrast; it is probably not valid to paraphrase as "This cat is black and that dog is black". It seems to me that "This is an amazing car" has the normal elaborator-center structure for "amazing car" and a normal equative predication of the subject, which should be reflected in the foundational layer—but there may be a focus on "amazing" in an information structure layer. The similarity of the two phrasings could be inferred from the foundational layer by the adjective's predicate-argument structure + the interpretation of the copula as equative, but there is a difference in the elaborators.

Does that make sense?

[jakpra]

What if the copular subject is clearly scene-evoking and the copular complement is more like an adverbial?

[The shock_P]_A [will]_F [be]_F [intense]_S or [The shock]_P [will]_F [be]_F [intense]_D

[Her_A response_P]_A [was]_F [short]_S or [Her]_A [response]_P [was]_F [short]_D

[jakpra]

Related: The "Expletive it" section in the guidelines presents the two examples:

1. “It_F is_F strange_S [that_R I_A met_P him_A here_A]_A”
2. “It_F is_F likely_S [to_R rain_P]_A”

I agree with the analysis of 1., but shouldn't 2. be It_F is_F likely_D to_F rain_P ?

Consider it in the following (more or less minimal) pairs, where my suggestion is much closer in structure to both alternative examples than the above "likely_S [to_R rain_P]_A" analysis: It_F is_F likely_D to_F rain_P John_A is_F likely_D to_F leave_P (p. 25)

It_F is_F likely_D to_F rain_P It_F is_F going_F to_F rain_P

[nschneid]

What if the copular subject is clearly scene-evoking and the copular complement is more like an adverbial?

My gut feeling is that D is better in these cases, but Omri/Dotan should weigh in.

Perhaps a good way to organize the section on copulas is with respect to whether the subject and complement evoke scenes, evoke semantic non-scene units, or are completely functional (pleonastic "it" subject). My hunch is that if both evoke semantic non-scene units, we should treat the copula as S with two participants; otherwise, the copula is F, and the scene-evoking subject or object is the main scene of the clause.

What about when both subject and object evoke full scenes (not adverbials)?

• (Inspired by the above paragraph): "My hunch is that he eats meat." is a way of saying "I suspect he eats meat.", or 'It is my hunch that he eats meat', so we could say the subject evokes the top-level scene: My_A hunch_S is_F [that_R he_A eats_P meat_A]_A

• "His meat-eating is a problem." This means 'It is a problem that he eats meat', not 'He eats meat problematically', so not problem_D. I guess the complement evokes the top-level scene: [His_A meat_A eating_P]_A is_F [a_E problem_C]_S

• "The wedding is a celebration of love": Here the copula is basically a light verb—it can be paraphrased as "The wedding celebrates love", with the complement evoking the top-level scene. [[The_E wedding_C]_P]_A is_F [a_E celebration_C]_P [of_R love_P]_A

[dotdv]

But a case like "this is an amazing car", when seen as a whole, seemed essentially much more similar to "this car is amazing" where it is easy to fill in the P/S spot with "amazing" as the S.

Ah, OK—I think there are two problems with this approach:

The practical problem is that it becomes awkward if there's more than one adjective. Consider "This is a nice red car and that is an ugly blue car." Under this policy I suppose you'd have 4 parallel scenes, whereas the syntax suggests 2 parallel scenes.

The theoretical problem is that "This is an amazing car" and "This car is amazing" may be interchangeable, but only in certain contexts with certain information structure. Consider "This is a black cat and that is a black dog", where it is the nouns that contrast; it is probably not valid to paraphrase as "This cat is black and that dog is black". It seems to me that "This is an amazing car" has the normal elaborator-center structure for "amazing car" and a normal equative predication of the subject, which should be reflected in the foundational layer—but there may be a focus on "amazing" in an information structure layer. The similarity of the two phrasings could be inferred from the foundational layer by the adjective's predicate-argument structure + the interpretation of the copula as equative, but there is a difference in the elaborators.

Does that make sense?

Yes, I understand, thanks. So according to the guidelines we annotate: John_A is_F amazing_S [amazing_S boy_A!]_H (we annotate this combination of adjective+noun directly as a scene only when it appears as a top layer unit, like we see a lot in the reviews)

But in the case of "John is an amazing boy" you suggest it be changed to: John_A is_F [an_E [amazing_S (boy)_A]_E boy_C]_A.

[dotdv]

What if the copular subject is clearly scene-evoking and the copular complement is more like an adverbial? My gut feeling is that D is better in these cases, but Omri/Dotan should weigh in.

Yes, we mark these Ds. a relevant example from the guidelines: His_A service_P was_F slow_D (p.21)

[omriabnd]

But in the case of "John is an amazing boy" you suggest it be changed to: John_A is_F [an_E [amazing_S (boy)_A]_E boy_C]_A.

Yes. I think so.

Dotan, could you integrate this issue into the "Classification of Copula/Verbless Clauses" section? I think the relevant section to change is this:

\item If both alienable and inalienable attributions appear in the same scene the alienable attribution will take precedence and we will mark it as the S:

\begin{itemize} \item ''This$_A$ is$_F$ [an$E$]${A-}$ amazing$_S$ [restaurant$C$]${-A}$'' \item ''This$_A$ is$_F$ [a$E$]${A-}$ red$_S$ [car$C$]${-A}$'' \end{itemize}

[dotdv]

But in the case of "John is an amazing boy" you suggest it be changed to: John_A is_F [an_E [amazing_S (boy)_A]_E boy_C]_A.

Yes. I think so.

Dotan, could you integrate this issue into the "Classification of Copula/Verbless Clauses" section? I think the relevant section to change is this:

\item If both alienable and inalienable attributions appear in the same scene the alienable attribution will take precedence and we will mark it as the S:

\begin{itemize} \item ''This$_A$ is$_F$ [an$E$]${A-}$ amazing$_S$ [restaurant$C$]${-A}$'' \item ''This$_A$ is$_F$ [a$E$]${A-}$ red$_S$ [car$C$]${-A}$'' \end{itemize}

@omriabnd I just noticed I made a mistake in the example you responded to: the discussion was on whether in such cases the copula should be marked S, but I accidently marked it F. So based on the discussion, it should've been: JohnA isS [an_E [amazing_S (boy)_A]_E boy_C]_A. If you agree with this I'll correct the relevant section.

[omriabnd]

Yes. Please do.

On Wed, Oct 24, 2018 at 9:47 AM dotdv notifications@github.com wrote:

But in the case of "John is an amazing boy" you suggest it be changed to: John_A is_F [an_E [amazing_S (boy)_A]_E boy_C]_A.

Yes. I think so.

Dotan, could you integrate this issue into the "Classification of Copula/Verbless Clauses" section? I think the relevant section to change is this:

\item If both alienable and inalienable attributions appear in the same scene the alienable attribution will take precedence and we will mark it as the S:

\begin{itemize} \item ''This$_A$ is$_F$ [an$E$]${A-}$ amazing$_S$ [restaurant$C$]${-A}$'' \item ''This$_A$ is$_F$ [a$E$]${A-}$ red$_S$ [car$C$]${-A}$'' \end{itemize}

@omriabnd https://github.com/omriabnd I just noticed I made a mistake in the example you responded to: the discussion was on whether in such cases the copula should be marked S, but I accidently marked it F. So based on the discussion, it should've been: JohnA isS [an_E [amazing_S (boy)_A]_E boy_C]_A. If you agree with this I'll correct the relevant section.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/UniversalConceptualCognitiveAnnotation/docs/issues/32#issuecomment-432533426, or mute the thread https://github.com/notifications/unsubscribe-auth/AIG86whjM1TVsR8_QFLyLU1a_NhjQ-ddks5uoAz5gaJpZM4W0o6x .

[dotdv]

OK, corrected that section

[dotdv]

Apart from the issue we addressed I see there are other questions here, so I labeled it both "addressed not reviewed" and "question"