Converb vs. Coverb in Mongolic and Turkic
dan-zeman opened this issue · comments
What is the difference between VerbForm=Conv and VerbForm=Coverb in Buryat? Similarly, Kazakh and Uyghur have VerbForm=Cov (in addition to Conv), which may possibly mean the same thing.
Skribnik (2011) lists over 10 different types of converbs for standard Buryat. One of them is the imperfective -zhA, which seems to match the "coverb" in our Buryat data: байжа.
Should we cover all these forms under VerbForm=Conv and then distinguish their subtypes in a language-specific feature?
Elena Skribnik. 2011. Buryat. In: The Mongolic Languages. Routledge
Probably VerbForm=Conv and VerbForm=Converb are the same.
VerbForm=Coverb is different, this is a form that is used with auxiliaries. In this article this is the equivalent of the "infinitive". But that label is not available in UD.
In general following the classification above:
- VN →
VerbForm=Vnoun - VADJ →
VerbForm=Part - VADV →
VerbForm=Conv - INF →
VerbForm=Cov
They are not intended to be the same; there are slight differences, and in my experience this is complicated by sometimes conflicting terminologies... even if, to me, such differences appear quite arbitrary or too much language-specific. Therefore, I am of the advice that all "adverbial forms" of verbs should be conflated into one single feature; other distinctions should happen through tense, mood, aspect, or syntactic relations.
If I remember well, one of the distinctions made in this field is if a verb in a given form can appear as a full predicate, or if it needs another "finite" verb; this involves also "participles". Boundaries are very blurry here, and traditional Western terminology does not help. Also, probably this distinction overlaps with that between regular and modal verbs in other languages, or also auxiliaries, as @ftyers points out, but then UD should already have all needed labels, no? This is surely complicated by the fact that at least in Mongolian nearly every word form seems to accept some kind of nominal inflection.
As an example, the descriptive grammar of Mongolian (not limited to Khalkha) by Janhunen (2012) does not use the term coverb, only converb, alongside participle. A main distinction is made between those converbs which admit different subjects than the main predicate ("disjunct") and those which don't ("conjunct"), among other of more morphological nature. There appears also the imperfective -ж/ч form you mention for Buryat.
So, there should be no need for coverbs, at least not in Mongolic! Just my 2 cents, let's call an Altaicist! 🙂
VerbForm=Coverb is different, this is a form that is used with auxiliaries. In this article this is the equivalent of the "infinitive". But that label is not available in UD.
Is the feature VerbForm=Inf not usable here?
In Turkish, infinitives are syntactically very similar to other forms of verbal nouns. So, we mark them with VerbForm=Vnoun. I expect this to be similar in other languages. Unless their distribution is (substantially) different than verbal nouns, my suggestion would be using Vnoun.
Turkish does not have "infinitives" of this type (unless you count e.g. the -A in -Abil). @jonorthwash any thoughts on this?
Now, this makes me wonder if all values for verbal nouns, like Vnoun, Inf or Sup (supine, which I don't know if it exists outside of Latin), shouldn't be just conflated into just one value. All these labels just seem to arise from language-specific grammatical traditions.
I mean, my point is that these objects are all substantially the same; if there are syntactic differences, then this should appear from relations and syntactic trees, rather than from labels, and morphology (presence/absence of declension, aspect, tense, mood...) is already marked by other features. In the same vein, I would also conflate Ger and Gdv into the more universal Part.
VerbForm=Coverbis different, this is a form that is used with auxiliaries. In this article this is the equivalent of the "infinitive".
Great, in that case we can use the existing VerbForm=Inf and we do not need a new label.
in Mongolian nearly every word form seems to accept some kind of nominal inflection
Can you point out a postposition, conjunction, or verb form other than a verbal noun or adjective that accepts nominal morphology in any Mongolic language?
Turkish does not have "infinitives" of this type (unless you count e.g. the
-Ain-Abil). @jonorthwash any thoughts on this?
There is an infinitive use of -(y)Ip with dur- (these are still written separately so are a slightly stronger case than the -(y)A forms you mention), e.g. in çay içip durdum.
@coltekin, the Turkish "-mAk infinitive" is a verbal noun by UD criteria. The term infinitive is reserved for verb forms that pattern with auxiliaries. (A case could be made with forms like çay içmek istiyorum, but I think this still could be a verbal noun, especially if it can answer the question ne istiyorsun?)
VerbForm=Coverbis different, this is a form that is used with auxiliaries. In this article this is the equivalent of the "infinitive".Great, in that case we can use the existing
VerbForm=Infand we do not need a new label.
It sounds like maybe you're merging converbs and infinitives? If so, I would express that these are very easily distinguished in Mongolic languages most of the time and probably shouldn't be merged in UD annotation.
Do I understand correctly that the forms in question are ones that require an auxiliary? If so, I see the point of Inf (and not Vnoun as in -mAk). The example from Turkish would be yap-a bilirim 'do-A abil-PRES-P1S' (I can do it), or iç-e durdum. I'd mark the -(I)p version, iç-ip durdum, as Conv, even though this specific example looks like iç-e durdum, its syntax is the same as other uses of -(I)p. These type of free auxiliaries are rare in Turkish (auxiliary becomes a suffix, as in yapabilirim, in typical use). In the first example above, it makes sense to mark the first verb yap-a as VerbForm=Inf. I think this also fits the definition/example in the UD documentation for Inf.
Now I wonder if the correct morphological tag for the non-finite forms like iç-iyor olacağım or iç-miş oldum should also be Inf. The first words (verbs) in these constructions do not have a particular suffix like -A above, but they are incomplete (as a finite or non-finite verb/head of a subordinate clause) without the auxiliary ol-.
I wouldn't say "require" an auxiliary, as there are no "infinitives" that are not also some other class (see Table 2 in this article).
These -iyor examples are interesting, and could be related to other copula-as-auxiliary constructions that we discuss in the paper.
Going with VerbForm=Inf for these "Coverbs" is fine by us (and corresponds to what we call them in the paper).
there are no "infinitives" that are not also some other class
... in Turkish.
These -iyor examples are interesting, and could be related to other copula-as-auxiliary constructions that we discuss in the paper.
They are exactly that.
I'd mark the -(I)p version, iç-ip durdum, as
Conv, even though this specific example looks like iç-e durdum, its syntax is the same as other uses of -(I)p
No, its syntax is entirely different. In verbal adverb (Conv) uses of -(y)Ip in Turkish, anything can come between the non-finite form and the main predicate. Also, in terms of the semantics, there are two separate events, not one as with the infinitive use. For example (feel free to correct details, but I think the point stands):
- Eve gidip, yeni kardeşlerimle Moğolistan'da aldığım çayı içtik.
I'm not positive about Turkish off the top of my head (and can't check easily as I'm typing this on my phone), but in other Turkic languages I know, the two verbs can even have different subjects.
In içip durdum, you can't put many things between them (maybe de and the like?), and they form a single predicate/event and must share a subject and all other arguments.
VerbForm=Coverbis different, this is a form that is used with auxiliaries. In this article this is the equivalent of the "infinitive".Great, in that case we can use the existing
VerbForm=Infand we do not need a new label.It sounds like maybe you're merging converbs and infinitives? If so, I would express that these are very easily distinguished in > Mongolic languages most of the time and probably shouldn't be merged in UD annotation.
No, I don't think so. The Buryat data currently has both VerbForm=Conv and VerbForm=Inf. The latter was previously labeled VerbForm=Coverb, the former is unchanged. But I cannot judge whether the distinction between the two is made correctly in the data.