Dative Subjects Saraiki about docs HOT 5 CLOSED

I am assuming that a subject (= agent) of a primary transitive predicate (such as X hits/hurts/kills Y) would not take the dative form, right? Nor would the dative be used for the object (= patient) of a primary transitive predicate? Then the dative nominal is not a core argument and does not qualify as a subject in the UD framework, despite the fact that it may be traditionally called "logical subject" or "dative subject" in non-UD approaches. The analysis with obl (or a subtype thereof) seems thus to be the correct one.

On the other hand, if تریۂ (tareh) "thirst" is in the nominative form and its gender / number / person is cross-referenced by the form of the verb, then it is the subject (nsubj) and not an object, regardless the semantic role. I think it is indeed the case — the glosses suggest that tareh is feminine and the verb is in feminine singular form (regardless of the gender of the experiencer), plus the auxiliary hai seems to be 3rd person singular (here I'm guessing based on similarity with Urdu/Hindi; I might be mistaken).

from docs.

Related old issue: #248

from docs.

Thanks,

yes, the subject of transitive verbs does not take the dative form and same for the objects.

Though the aux is same as in Urdu and Hindi, but it is not agreeing with the subject but providing the tense information (present in the given sentence). Based on my understanding of your comment it should be like :

1	میکوں	_	PRON	_	_	3	obl:agent	_	_
2	تریۂ	_	NOUN	_	_	3	nsubj	_	_
3	لگی	_	VERB	_	_	0	root	_	_
4	ہے	_	AUX	_	_	3	aux	_	_

Please correct me if I misunderstood.

from docs.

... Based on my understanding of your comment it should be like :

1 میکوں _ PRON _ _ 3 obl:agent _ _ 2 تریۂ _ NOUN _ _ 3 nsubj _ _ 3 لگی _ VERB _ _ 0 root _ _ 4 ہے _ AUX _ _ 3 aux _ _

Please correct me if I misunderstood.

Yes, this is how I would annotate the sentence – perhaps I would be just hesitant about the ":agent" in obl:agent, as it suggests a different semantic role than experiencer. But then we admit the same inaccuracy with passives (where the oblique "agent" does not necessarily have the proto-agent role), so it is probably OK here, too.

from docs.

Thank you :)

from docs.