Nepali - how to handle plural morpheme on verbs to mark politeness for singular person

[geetsgithub]

Hello,
In Nepali, the verbal morpheme for marking plural number can also be used for marking politeness for third-person singular subjects. In this case, should we use the Polite feature alongside the Number feature, and annotate the morphological form that agrees with the logical number of the subject? For example, If the verb or modifier is in the plural form to show respect to a single person, should we mark it as Number=Sing (or should we underspecify the Number feature?) and add the politeness feature?

Thank you.

[Stormur]

If it is the same marker, the annotation should stick to its "basic" value and not annotate the context. A discrepancy between grammatical and semantic number is actually interesting to observe.

Then to that you add the Polite feature, which with some values, I understand, is a way to give the context of use (even if it also encompasses forms which are specifically for politeness, and nothing else). Incidentally, I fear that this feature is not used consistently across treebanks.

[geetsgithub]

When you say "If it is the same marker, the annotation should stick to its "basic" value and not annotate the context," do you mean I should annotate the verb with the Number=Plur feature even if it does not logically agree with the subject number (which is singular)? I plan to use the feature Polite=Form to indicate "formal" politeness. When its neutral, I don't add Polite feature.

[geetsgithub]

In UD, it seems our goal is to annotate sentences based on the surface-level form of words (not based on contexts). I have another similar situation. In Nepali, a noun can functionally be plural, even without a plural suffix (which can be optional), and its plurality is given by the context (e..g, when the verb conjugation indicates the noun is plural). In such situations, do annotate the noun based on its syntactic behavior and verb agreement (Number=Plu), or based on the surface level form, which will be Number=Sing?

[sylvainkahane]

In French we have a similar problem: the 2PL pronoun vous is used as a polite form. Interestingly it triggers a plural agreement of the verb but a singular agreement of the adjective: vous êtes normal 'you be.PL normal.SG'. Conversely, the 1PL pronoun nous has been replaced by the 3SG pronoun on (same root as English one), which triggers a 3SG agreement of the verb and plural agreement of the adjective: (nous,) on est normaux lit. (us,) one is normal.PL = 'we are normal'.
We propose to use a double marking of the pronoun:

vous: Person=2, Number=Plur, Number[sem]=Sing, Polite=Form
on: Person=3, Number=Sing, Person[sem]=1, Number[sem]=Plur.

[Stormur]

When you say "If it is the same marker, the annotation should stick to its "basic" value and not annotate the context," do you mean I should annotate the verb with the Number=Plur feature even if it does not logically agree with the subject number (which is singular)? I plan to use the feature Polite=Form to indicate "formal" politeness. When its neutral, I don't add Polite feature.

Yes, exactly. Just the fact that you say that "the plural is used for politeness" (like, e.g. in German with 3rd person, or in old-fashioned Italian and French with 2nd person) is already giving directions on how to annotate.

My further question is whether this disalignment is only logical, so given the context of using a plural with respect to an individual referent, or whether it also appears in syntax, with something like "that person are nice".

[Stormur]

In UD, it seems our goal is to annotate sentences based on the surface-level form of words (not based on contexts). I have another similar situation. In Nepali, a noun can functionally be plural, even without a plural suffix (which can be optional), and its plurality is given by the context (e..g, when the verb conjugation indicates the noun is plural). In such situations, do annotate the noun based on its syntactic behavior and verb agreement (Number=Plu), or based on the surface level form, which will be Number=Sing?

Exactly, if you want the plural marker is the "surface" that is annotated.

Absolutely, in those cases I would advise against using a contextually determined Number=Plur. Actually, you could even decide that Nepali "basic forms" are not marked for Number in any way, and so not annotate Number unless it is explicitly marked (as plural).

Another way to see it: can you retrieve the Number of the noun only by looking at the verb form, and not at the noun form? If this is the case, then no annotation of Number goes on the noun, only on the verb.

[amir-zeldes]

In basic UD for English (without extensions like [sem]), we use form based annotations where the form is clearly wrong and add Typo and CorrectForm to indicate the correct form like this:

• type, Number=Sing -> CorrectForm: types

1	There	there	PRON	EX	PronType=Dem	2	expl	2:expl	Discourse=context-background:11->14:1:sem-synym-108,125-126-_
2	are	be	VERB	VBP	Mood=Ind|Number=Plur|Person=3|Tense=Pres|VerbForm=Fin	0	root	0:root	Cxn=Existential-CopPred-ThereExpl
3	different	different	ADJ	JJ	Degree=Pos	4	amod	4:amod	Entity=(31-abstract-new-nnnnn-cf3-2-sgl|MSeg=differ-ent
4	type	type	NOUN	NN	Number=Sing|Typo=Yes	2	nsubj	2:nsubj	CorrectForm=types|CxnElt=2:Existential-CopPred-ThereExpl.Pivot|XML=<sic ana:::"types"></sic>
5	of	of	ADP	IN	_	7	case	7:case	_
6	lighting	lighting	NOUN	NN	Number=Sing	7	compound	7:compound	Entity=(32-object-new-nnnnn-cf5-2-sgl|MSeg=light-ing
7	bulbs	bulb	NOUN	NNS	Number=Plur	4	nmod	4:nmod:of	MSeg=bulb-s

But in ambiguous situations, where the form is not simply incorrect, we do use context to disambiguate number. For example "data" is Sing unless the verb has plural agreement, in which case we use Plur:

• data (show), Number=Plur

1	Our	our	PRON	PRP$	Case=Gen|Number=Plur|Person=1|Poss=Yes|PronType=Prs	2	nmod:poss	2:nmod:poss	Discourse=attribution-positive:24->25:0:sem-atsrc-256-257-_+lex-indwd-258-_+syn-rpr-259-_|Entity=(53-abstract-new-nnnnn-cf9-2-sgl(2-person-giv:inact-nsnsn-cf3-1-ana)|PDTB=Implicit:Expansion.Conjunction:and:_:221-255:256-281
2	data	data	NOUN	NNS	Number=Plur	3	nsubj	3:nsubj	Entity=53)|MSeg=dat-a
3	show	show	VERB	VBP	Mood=Ind|Number=Plur|Person=3|Tense=Pres|VerbForm=Fin	0	root	0:root	_
4	that	that	SCONJ	IN	_	16	mark	16:mark	Discourse=elaboration-additional:25->6:5:sem-rpt-49-50,276-277-_+sem-rpt-76-77,266-267-_

• data (was), Number=Sing

1	It	it	PRON	PRP	Case=Nom|Gender=Neut|Number=Sing|Person=3|PronType=Prs	2	nsubj	2:nsubj	Discourse=joint-sequence_m:18->10:1:_|Entity=(23-abstract-new-nnnnn-cf3-2-disc
2	turned	turn	VERB	VBD	Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin	0	root	0:root	MSeg=turn-ed
3	out	out	ADP	RP	_	2	compound:prt	2:compound:prt	_
4	the	the	DET	DT	Definite=Def|PronType=Art	7	det	7:det	Entity=(3-abstract-giv:inact-sssss-cf4-4-coref
5	"	"	PUNCT	``	_	7	punct	7:punct	SpaceAfter=No|XML=<q>
6	important	important	ADJ	JJ	Degree=Pos	7	amod	7:amod	MSeg=import-ant
7	data	data	NOUN	NN	Number=Sing	13	nsubj	13:nsubj	SpaceAfter=No
8	"	"	PUNCT	''	_	7	punct	7:punct	Entity=3)|XML=</q>
9	was	be	AUX	VBD	Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin	13	cop	13:cop	_
10	her	her	PRON	PRP$	Case=Gen|Gender=Fem|Number=Sing|Person=3|Poss=Yes|PronType=Prs	13	nmod:poss	13:nmod:poss	Entity=(3-abstract-giv:act-sssss-cf4-3,4-coref(2-person-giv:inact-sssss-cf1-1-ana)
11	Candy	Candy	PROPN	NNP	Number=Sing	13	compound	13:compound	Entity=(24-abstract-new-sssss-cf5-1,2-coref-Candy_Crush_Saga
12	Crush	Crush	PROPN	NNP	Number=Sing	11	flat	11:flat	Entity=24)
13	progress	progress	NOUN	NN	Number=Sing	2	ccomp	2:ccomp	Entity=3)23)|SpaceAfter=No

[Stormur]

English actually very much looks a language where Number is only annotated positively for the plural, and very often semantically determined with respect to agreement for the rest.

data is one example, then uses such as police have... or Manchester United have scored, some mass nouns, etc.

[geetsgithub]

When you say "If it is the same marker, the annotation should stick to its "basic" value and not annotate the context," do you mean I should annotate the verb with the Number=Plur feature even if it does not logically agree with the subject number (which is singular)? I plan to use the feature Polite=Form to indicate "formal" politeness. When its neutral, I don't add Polite feature.

Yes, exactly. Just the fact that you say that "the plural is used for politeness" (like, e.g. in German with 3rd person, or in old-fashioned Italian and French with 2nd person) is already giving directions on how to annotate.

My further question is whether this disalignment is only logical, so given the context of using a plural with respect to an individual referent, or whether it also appears in syntax, with something like "that person are nice".

If a sentence "that person are nice" is said with honorific verb conjugation, the demonstrative "that" has to be changed to "those" and the adjective 'nice' also gets marked with its plural morpheme. Only the subject remains singular. So it would be something like "those person are nice" .

[Stormur]

If a sentence "that person are nice" is said with honorific verb conjugation, the demonstrative "that" has to be changed to "those" and the adjective 'nice' also gets marked with its plural morpheme. Only the subject remains singular. So it would be something like "those person are nice" .

Is this because in general a noun like "person" will not necessarily be marked for plural, also in other contexts?

As previously discussed, all this picture anyway points to a regular annotation with Number=Plur where it is marked, no Number where it is not (like on the noun "person" in the example), and possibly the feature Polite somewhere.

Personally I would see Polite in the MISC field, or with some layer (e.g. Polite[context]) as it does not correspond to anything morphological really. Also, it should be annotated on each and every word in the sentence... at that point I am not sure it makes sense anymore, at least not in similar cases, and instead some comment to the whole sentence could be useful to give the context.

[geetsgithub]

Is this because in general a noun like "person" will not necessarily be marked for plural, also in other contexts?

If the noun is marked with a plural suffix in this context, it cannot refer to a single person. Then the sentence will mean "those people are nice". When a plural demonstrative is used, we need to use a plural suffix, if we are referring to plural nouns.