Veni, vidi, vici - parataxis or conj?

[Stormur]

As the title suggests, I found a contradiction in the documentation regarding the treatment of the Latin sentence

Veni, vidi, vici
I came, I saw, I won

In the documentation, it is listed as an example of asyndetic co-ordination, so deprels are conj there; in slide 26 of the UD tutorial it is used as example of side-by-side sentences, so of parataxis.

What is the correct interpretation? Reading the documentation, I would actually be inclined towards parataxis, but the border between the two cases, when predicates are co-ordinated, looks really blurred to me. Is it thinkable to restrict asyndetic co-ordination to non-verbs only?

[amir-zeldes]

+1 for parataxis here, but I think you can also have conj for verbs with zero VP coordination, i.e. if they share a subject:

She went to work, made some coffee.

Here conj would let you recover the subject deterministically, but parataxis wouldn't. In the Latin pro-drop cases this doesn't really make a difference, but if there's a shared subject I would go with conj (that's what we did in UD_English-GUM anyway).

[nschneid]

To me the clearest case of conj without a coordinating conjunction is seen in headlinese:

Budget's oil provisions divide Congress, White House

The comma is just a terse substitute for "and". A full sentence paraphrase might be The budget's oil provisions divide Congress and the White House.

[dan-zeman]

I would not restrict asyndetic coordination to non-verbs (and I would not restrict parataxis to verbs).

This boundary indeed is blurred and I suspect that different UD'ers have different opinions here but unlike some of the more prominent issues, we haven't tried to sort it out and come up with more detailed guidelines.

I am personally not fond of parataxis competing with conj. I would always prefer conj if the construction remotely resembles the more prototypical examples of coordination; therefore, veni, vidi, vici is coordination for me.

[nschneid]

I would always prefer conj if the construction remotely resembles the more prototypical examples of coordination; therefore, veni, vidi, vici is coordination for me.

I think I have the opposite instinct: for me, the so-called "comma splice" is prototypical parataxis:

I lost my wallet, hopefully somebody will find it.

This can be paraphrased either as separate sentences or with a coordinating conjunction.

@dan-zeman, would you think conj could apply there? If so, maybe we should consider renaming parataxis to parenthetical.

[amir-zeldes]

I agree with @nschneid and I think the example with Congress + White House is also conj because it is also a case of structure sharing (they're part of the same argument structure). In the wallet example, both clauses are fully independent.

[dan-zeman]

@nschneid : Yes I would think conj could apply there and I believe we would find such examples of conj in UD Czech because it was converted from data where such things are annotated as coordination. (And because the source corpus distinguishes parentheticals, but not parataxis in the broader UD sense.)

But I'm also pretty sure that other people who participated in shaping the guidelines would disagree with me. Definitely do not take my opinion as the "official" position on this.

Renaming parataxis to parenthetical, with the appropriate narrowing down of the definition, would have to wait for UD v3 guidelines, but I think I'd support it.

[sylvainkahane]

We have to face this problem everytime when we work with spoken copora. You have no punctuation and you have to decide when a sentence ends or not. We use the following criteria for coordination:

A coordination occurs only in a governed position. No governor, no coordination.

I think it is a good criteria, easy to apply.

Example: Veni, vidi, vici. No governor, I can't use conj. So I use parataxis. If it was a spoken data, I would have considered that I have three sentence and three roots.

Note also that parataxis has many uses and in our treebank of spoken French, we use several subtypes of parataxis, but it is another topic. In this case we use parataxis:conj.

[dseddah]

Le 12 avr. 2018 à 18:00, sylvainkahane ***@***.***> a écrit : We have to face this problem everytime when we work with spoken copora. You have no punctuation and you have to decide when a sentence ends or not. We use the following criteria for coordination: A coordination occurs only in a governed position. No governor, no coordination. I think it is a good criteria, easy to apply. Example: Veni, vidi, vici. No governor, I can't use conj. So I use parataxis. If it was a spoken data, I would have considered that I have three sentence and three roots. Note also that parataxis has many uses and in our treebank of spoken French, we use several subtypes of parataxis, but it is another topic. In this case we use parataxis:conj.

talking about parataxis, would you make the same analysis for No. Way! ? or IM. POS. SI. BLE ! without the period I would have gone with goes_with for the second case but here it could be parataxis:goes_with (if it doesn’t exist yet, I’m voting for it) Djamé

[nschneid]

No. Way!

IM. POS. SI. BLE !

I would treat the punctuation here as only adding stylistic/pronunciation emphasis, not changing the morphological or syntactic structure. These don't seem like parataxis to me.