complements of "be"

[jonorthwash]

In various Turkic treebanks, the verb ol- / бол- "be" is treated as a VERB, instead of AUX for copula constructions (see e.g. note at bottom of https://universaldependencies.org/tr/dep/cop.html).

There are a few reasons it was done this way, which I'm happy to get into.

But then how should complements of be verbs be treated?

E.g. Deniz doktor olacak. "Deniz will be a doctor."

Is doktor xcomp or obj of olacak "will be"?

In UD_Turkish-BOUN these nominal complements are treated as obj. Adjectival complemenets of ol- are sometimes treated as amod (which definitely doesn't seem right to me) and sometimes as obj.

[jonorthwash]

A few examples from UD_Turkish-BOUN:

22  gerçek  gerçek  ADJ Adj _   23  obj _   _
23  olmayan ol  VERB    Ptcp    Polarity=Neg|Tense=Pres|VerbForm=Part   24  acl _   _

gerçek means "true, real"

12  doğru   doğru   ADJ Adj _   13  amod    _   _
13  olmadığını  ol  VERB    Ptcp    Aspect=Perf|Case=Acc|Number=Sing|Number[psor]=Sing|Person=3|Person[psor]=3|Polarity=Neg|Tense=Past|VerbForm=Part    8   conj    _   _

doğru means "correct"

[dan-zeman]

amod is incorrect, even if doğru means "correct" :-)

obj under something that is almost a copula does not seem right either.

We normally recommend xcomp for dependents of "to become" and other pseudo-copulas (secondary predication). So I would go with xcomp here, too.

[jonorthwash]

We normally recommend xcomp for dependents of "to become" and other pseudo-copulas (secondary predication).

This is actually one of the reasons we annotate these verbs as VERB and avoid cop relations: they usually also mean "become".

So I would go with xcomp here, too.

Thanks, that's what I was thinking, but I couldn't find anything in the docs.

[jonorthwash]

amod is incorrect, even if doğru means "correct" :-)

obj under something that is almost a copula does not seem right either.

Should an issue be filed against UD_Turkish-BOUN?

[dan-zeman]

amod is incorrect, even if doğru means "correct" :-) obj under something that is almost a copula does not seem right either.

Should an issue be filed against UD_Turkish-BOUN?

Yes, please.

[Stormur]

We normally recommend xcomp for dependents of "to become" and other pseudo-copulas (secondary predication).

This is actually one of the reasons we annotate these verbs as VERB and avoid cop relations: they usually also mean "become".

Just a side-question: is this meaning of "become" tied to particular TAME combinations?

[jonorthwash]

Just a side-question: is this meaning of "become" tied to particular TAME combinations?

No, not really. In fact, the example sentence above, Deniz doktor olacak, could be read "Deniz will be a doctor" or "Deniz will become a doctor". Likewise, Deniz doktor oldu could be read "Deniz was a doctor" or "Deniz became a doctor." I think.

In a Turkic language with more explicit aspect marking (e.g., Central Asian Turkic languages, which use auxiliaries to do this, and even some have different morphology e.g. for imperfects), I think you do get some differences. E.g., in Kyrgyz Дениз доктур болуп калды would mean "Deniz became a doctor" and Дениз доктур болчу would mean "Deniz was / used to be a doctor", without any ambiguity in translation between "become" and "be".

[jonorthwash]

Likewise, Deniz doktor oldu could be read "Deniz was a doctor" or "Deniz became a doctor." I think.

Perhaps a native speaker doesn't get the "Deniz was a doctor" reading. This might be better expressed by Deniz doktordu (with a defective copula verb, not the ol- verb). I defer to native speakers.

[jonorthwash]

amod is incorrect, even if doğru means "correct" :-) obj under something that is almost a copula does not seem right either.

Should an issue be filed against UD_Turkish-BOUN?

Yes, please.

Done.