xcomp in Swedish

[AleksandrsBerdicevskis]

As far as I can see, in constructions as hjälpa någon att VERBa, VERBa consistently gets an advcl in Talbanken and an xcomp in LinES, and I think xcomp is the right choice.

In LinES, however, I find the following example: Den studentikosa form av självförverkligande, som brukar komma till uttryck när engelsmän bestämmer sig för att ge livslusten fria tyglar, dominerade en stund., where ge is an advcl, while I think it should be an xcomp.

(I thought we saw more cases of potentially missing xcomps, but I can't find them now. Perhaps @harisont remembers more.)

[jnivre]

Sounds reasonable. I will put this on my todo list.

[harisont]

Right, I am under the impression that advcl is slightly overused (not always at the expense of xcomp though), but unfortunately I did not keep track of the examples I found suspicious. If I stumble upon them again, I'll make them known.

[jnivre]

Please do. Thanks!

[harisont]

Another case is "tänka på att X". In LinES, this is always treated as ccomp/xcomp, in Talbanken I see only advcl (but "fundera på att X" is xcomp even in Talbanken (edit: sorry, that was still LinES)).

[jnivre]

This one is trickier (than #1132) because here I think it is clear that "på" is a preposition, not a particle. Hence, in the nominal case ("tänka på det"), it has to be "obl", not "obj". By consequence, it has to be "advcl", not "ccomp/xcomp" in the clausal case, because it is not a core argument.

This is a bit of a quirk in the UD guidelines, and @amir-zeldes has often argued that there should be a "cobl" dependency, for the clausal equivalent of "obl", so that we won't be forced to lump these together with normal adverbial clauses. On the other hand, we do lump oblique arguments (like "tänka på det" = "think of that") with prepositional adverbials (like "tänka i en timme" = "think for an hour"), so it is a moot point.

But with the current UD inventory of relations and guidelines for the core-oblique distinction, these have to be "advcl".

[harisont]

Yes, I defninitely see your point, but does that mean that most of the cases where and ADP is labelled as mark of a *comp need to be reviewed (or at least, all cases in LinES where "på att" introduces a *comp)?

A couple examples from the latter group where I'd find advcl particularly perplexing:

• Efteråt visar historiska studier på att det faktiska förloppet saknat varje spår av den sortens intelligens
• Han hade varit säker på att poolskötarna hade ögonen på honom

[jnivre]

Yes, I'm afraid that these have to be "advcl" under the current guidelines.

The thing to keep in mind is that UD does not distinguish complements vs. adjuncts, only core arguments vs. oblique dependents, so it is all about the morphosyntactic encoding (in this case whether there is a "flag" or not), not about valency. Admittedly, this is not well described in the current guidelines, so maybe the guidelines group should make a clarification about this. I can bring it up at our next meeting.

[AleksandrsBerdicevskis]

The thing to keep in mind is that UD does not distinguish complements vs. adjuncts, only core arguments vs. oblique dependents, so it is all about the morphosyntactic encoding (in this case whether there is a "flag" or not), not about valency. Admittedly, this is not well described in the current guidelines, so maybe the guidelines group should make a clarification about this. I can bring it up at our next meeting.

That'd be great! I remember wondering several times whether UD cares about valency or not.