search

tibetan-nlp / annotation-docs

Tibetan annotation docs
1 stars 0 forks source link

Need to establish guidelines for tricky arg1 dependencies #20

Open heacu opened 6 years ago

heacu commented 6 years ago

Consider the following example (see original):

020b

He was conscious of his own great strength and remained there without fear.

Is kho really the arg1 of yod pa? I suspect not, and the clue is the case marking: we have kho rather than khor / kho la, which suggests that kho is probably the subject of bzhag, not yod.

samyorode commented 6 years ago

Of course, one could argue that yod pa has a missing subject which probably would be marked with la or -r, but I am not sure if that would always be the case. Let's see what further annotation will bring to light. So kho is subject of bzhag here!

bildschirmfoto 2018-01-09 um 12 51 39
heacu commented 6 years ago

I agree that we can't take for granted that the subject of yod pa must be marked with la or -r. Moreover, if it must be, then that's something we should like to establish by corpus statistics and not just by assumption. So I'll accept the change but keep this issue open, in case other similar examples pop up that might help us.