"inner" vs. "outer" structure, spec. for poss. & clausal structures

[IsraelLand]

Hi @amir-zeldes

It seems we prefer the "external" analysis (not sure of the correct phrasing), as in, we tag ואת האיש ההוא and obj.marker that man

as conj, while only the "et" and its case=Acc suggest there's an inward object structure. Same with appos. and other deprels.

I've lately encountered stuff like - בית מ שלו a house of his [own]

שכר מופחת מהמגיע להן (= מופחת מהשכר המגיע להן) a salary, reduced of [the salary\what] that they deserve

I assume both to get the "parent analysis", meaning nmod, not nmod:poss, obl not advcl, respectively. If so, we need to implement it further in validation. I think this is consistent with the broader tendency of external\outer tagging, wdyt? Thank you

[IsraelLand]

a slightly related topic is anything similar to - הארכה מעבר לנהוג כיום an extention beyond the accepted today

Since "nahug" is adjectival (extension is a local nominal head as well), this may warrant advcl instead of nmod, unless again, we do the external\outer analysis.

The more crucial question is not this sentence, really, but in regards to the "newly-unfixed" adjectical stuff like - מצאתי אותם בנוסף ל.... אובחנו נכון להיום found them in addition to... were diagnosed as of (lit. right of) today which wouldn't be obl, but advcl, then (as we ditched the fixed analysis) if we don't tag by the external.

[amir-zeldes]

In general I think it's exactly as you say, and you will find the case-sisters analysis in English and other languages for things like משלהן etc., so the external deprel is determined by the outermost preposition. This also makes sense, since in fact it doesn't mean "his house", it means "a house of his own", so the possession is 'wrapped' in another PP.

About שכר מופחת מהמגיע להן, this is complicated by the possibility of using ha- for she- in Hebrew, so I actually think advcl is possible here, but then indeed you would have to treat ha- as mark and SCONJ, which is maybe fine actually. That would fit the 'nahug' case as well.

[IsraelLand]

Thank you. following the advcl \"cobl" logic,wouldn't it mean a different analysis for the no longer fixed adjectivals?

They used to be fixed, so for example (assuming all these are either adjectival or verbal) -

מצאתי בנוסף לנהוג [ש...]

advcl(matzati, nahug) mark/case(nahug, be nosaf le) fixed(be, nosaf, le)

But afaik changed to the current -

obl(matzati,nosaf) obl(nosaf,nahug) case(nosaf,be)

So I'm wondering if the above obl's shouldn't be advcls, if they're adjectival. Very inconvenient, but perhaps warranted by the definition of advcl/cobl.

Even if we don't treat "nosaf" as an adjective, I see no way to avoid that with אובחנו נכון להיום, so an advcl(uvxenu,naxon), rather than obl:npmod we utilise today.

Don't get me wrong, I'm very much for the current tagging of obl, but just been wondering as to the extent of the cobl aspect of advcl. Thank you

[IsraelLand]

Also about the שכר מופחת מהמגיע להן - would the "me" adp "turn into" sconj, or stay its original adp? On the one hand, on the parent side it's adp, but if we look at the cobl/advcl side (on its first level - "me", not "ha") it might need to conform to it being a part of an advcl (again regardless of the "ha", so for example עייפים מלעשות) Thanks

[amir-zeldes]

I think it's a single decision: if it's advcl, then you also use SCONJ. If you use ADP, that means you think "מגיע" is nominalized. I think SCONJ+advcl makes sense here, but it's open to debate.

[IsraelLand]

Yeah I see where you're coming from. Thanks