You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
"such"/det licensing an advcl, as in these results. The guidelines on sufficiency and excess for "so" and similar say the advcl should attach to the adjective or adverb, not the noun in a case like sufficient flour. In such a high price that nobody could afford it, I suppose "such" should have an advcl dependent?
English treebanks distinguish 2 uses of "such": an ADJ one in most contexts, and a DET specific to when it occurs before an indefinite article. EWT trees as of the 2.15 release:
ADJ (amod)
DET (det:predet)
Note the in both cases, "such" heads an advcl which is the clause licensed by the causal excess construction.
Compare "enough flour to get started" in the guidelines, where "enough" is an ADJ and heads an advcl.
Despite these two uses of "such" looking very similar, in the 2.16 release we cannot maintain the structure in the DET case because the validator is now enforcing a requirement that det words not have advcl dependents.
So it seems we have two options:
Prioritize the DET/det:predet analysis of the predeterminer (which follows the Penn Treebank tag PDT, though it is somewhat fraught: "quite" as predeterminer UD_English-EWT#412), and attach the subordinate clause under the noun as an acl when "such" is a predeterminer:
Decide that "such" should uniformly be an ADJ for UD purposes, so the structures can be parallel:
Treating "such" in predeterminer position as an adjective is actually more consistent with the Huddleston & Pullum analysis, but it would be a departure from PTB in an edge case.
In the interest of stability (det:predet in direct correspondence with Penn Treebank PDT), @amir-zeldes prefers Option 1. The rest of the Core Group was open to Option 2 but not expressing a strong preference. I personally do not like the PTB/current approach to det:predet but that is a larger conversation and we need to decide a solution now for the 2.16 release. So I will implement and document Option 1. Note that it affects just 4 instances of predeterminer "such" + complement clause in EWT.
Originally raised in #1059:
English treebanks distinguish 2 uses of "such": an ADJ one in most contexts, and a DET specific to when it occurs before an indefinite article. EWT trees as of the 2.15 release:
ADJ (amod)
DET (det:predet)

Note the in both cases, "such" heads an
advclwhich is the clause licensed by the causal excess construction.Compare "enough flour to get started" in the guidelines, where "enough" is an ADJ and heads an
advcl.Despite these two uses of "such" looking very similar, in the 2.16 release we cannot maintain the structure in the DET case because the validator is now enforcing a requirement that
detwords not haveadvcldependents.So it seems we have two options:
aclwhen "such" is a predeterminer:Treating "such" in predeterminer position as an adjective is actually more consistent with the Huddleston & Pullum analysis, but it would be a departure from PTB in an edge case.
In the interest of stability (
det:predetin direct correspondence with Penn Treebank PDT), @amir-zeldes prefers Option 1. The rest of the Core Group was open to Option 2 but not expressing a strong preference. I personally do not like the PTB/current approach todet:predetbut that is a larger conversation and we need to decide a solution now for the 2.16 release. So I will implement and document Option 1. Note that it affects just 4 instances of predeterminer "such" + complement clause in EWT.