NUM missing NumForm/NumType

[nschneid]

https://universal.grew.fr/?custom=653d1ce18d128

(Depends on #464)

[rhdunn]

In UniversalDependencies/UD_English-GENTLE#5, I've proposed a NumForm=Alpha feature for the (a) etc. cases, as they don't currently have a sensible value to use as they are not words.

[nschneid]

The guidelines aren't clear about how to tag letters functioning like numbers (marking items sequentially). I think that needs to be resolved before deciding whether such items deserve a NumType. This should be raised at https://github.com/UniversalDependencies/docs/issues.

[rhdunn]

I've raised UniversalDependencies/docs#983 for the discussion on alphabetic list forms.

[amir-zeldes]

I'm with @dan-zeman that these are not really numbers, so I don't think they should have NumType. There are languages that truly use letters as numbers, mostly ancient languages (Coptic, Biblical Hebrew, Ancient Greek etc.). In those cases I would see the case for a NumType like this, but not for English list item markers. More in the other thread.

[nschneid]

The question has arisen: assuming list enumerators like "(A)" are NUM, what should their features be?

• NumForm: My hunch would be to add a new value NumForm=Letter (because it's not a spelled-out word in the same way that "one" is).
• NumType is harder to decide because we don't normally have an ordinal/non-ordinal contrast ("the cth item"), but if Ord is construed narrowly as requiring a special suffix, then Card fits I guess? I don't think it should really be a semantic feature—morphologically speaking, "1" fits NumType=Card regardless of whether it has a meaning of quantity ("1 book") or counting ("item number 1", "1. an item" etc.). If list item markers like "1" or "(1)" can be NumType=Card then I guess that can be extended to "(A)".

(I am setting aside German spelling, where I take it "1." can be used within a sentence to mean 'first', so the period makes it explicitly ordinal.)