Wikidata:Property proposal/Grammatical Person

grammatical person

Originally proposed at Wikidata:Property proposal/Lexemes

   Done: grammatical person (P11053) (Talk and documentation)

Description	Grammatical person category (first, second, third, etc.) of pronouns
Represents	grammatical person (Q690940)
Data type	Item
Domain	Lexemes with grammatical category personal pronoun (Q468801) or pronoun (Q36224)
Allowed values	first person (Q21714344), second person (Q51929049), third person (Q51929074)
Example 1	I (L487) → first person (Q21714344)
Example 2	you (L482) → second person (Q51929049)
Example 3	he (L485) → third person (Q51929074)
Example 4	she (L484) → third person (Q51929074)
Source	https://www.smg.surrey.ac.uk/features/morphosyntactic/person/

Motivation

Grammatical properties which apply to a lexeme as a whole (i.e. to all its forms) should be stated on the lexeme, rather than repeated on every form. Thus, according to this logic, personal pronouns which have a fixed grammatical person feature should state it on the lexeme level through this property.

Please note that the grammatical person property should strictly only refer to the person feature and not any other features (gender, number, etc.) per this discussion.

Note that this property has been suggested before but rejected on the false grounds that this is not a property of lexemes but rather of forms. However, as stated above, it is a property of personal pronouns, insofar they exhibit a fixed grammatical person feature. AGutman-WMF (talk) 14:54, 17 June 2022 (UTC)

Update: Please note that I have created a complementary request for the "grammatical number" property (needed as well for pronouns). AGutman-WMF (talk) 11:50, 10 July 2022 (UTC)

Discussion

• @Tubezlob: as the proposer of the original proposal, @Duesentrieb, Deryck Chan: as the purveyors of "false grounds", and @JakobVoss, VIGNERON: as those who were uncertain last time. Mahir256 (talk) 17:19, 17 June 2022 (UTC)
• Still not entirely sure but leaning towards  Support if it's limited to pronouns. Cheers, VIGNERON (talk) 19:27, 17 June 2022 (UTC)
• @Mahir256: Thanks for the high praise. I don't have any translingual "false ground" to purvey this time - all the languages I speak can be neatly analysed in the lenses of first, second, and third person syntactically. But I have two questions on the implementation of this proposal:
  1. @AGutman-WMF: This property proposal seems redundant to the "Lexical Category" built-in feature of Lexemes. For example he (L485) F1-F4 already link to third person (Q51929074). What use cases do you have in mind that isn't redundant to Lexical Category?
  2. @VIGNERON: Why should this be limited to pronouns? In many fusional languages, grammatical person is expressed through verb conjugations.

  So on balance, weak oppose for now until we agree on these two issues. Deryck Chan (talk) 14:29, 18 June 2022 (UTC)

  @Deryck Chan: there is two things here: person of a form and person of a lexeme. he (L485) is a great example: why put the gender in both lexeme and forms (each and every time the same since it's a characteritic of the lexeme) and the person only in forms? Shouldn't it both be stored only at the lexeme level? (precisely in order to avoid redundancy)

  Pronouns are the only lexical category (I know of) where the person is a characteritic of the lexeme and not a flexion. It reminds me of the gender in some european language, for nouns the gender is a characteritic of the lexeme (le soleil soleil (L11620), die Sonne Sonne (L6775)) but for other categories like adjectives or verbs, the gender is characteritic of the form (amicaux/amicales amical (L624418)).

  Cheers, VIGNERON (talk) 15:07, 18 June 2022 (UTC)

  I'm not sure how the Lexical Category helps us here. he (L485) has personal pronoun (Q468801) as lexical category, but that doesn't inform us on the grammatical person. third person (Q51929074) is indeed linked to the individual forms, but my point is exactly that whenever a grammatical feature is given on all forms it should be elevated to a lexeme-level statement. For lexemes whose forms inflect for the person category, e.g. verbs, the person feature must be listed individually on each verbal form. If we were to include individual verbal suffixes or prefixes in Wikidata, then they may be similar to stand-alone pronouns in that they could have a fixed person category. AGutman-WMF (talk) 09:32, 20 June 2022 (UTC)

  • Makes sense to me, thank you AGutman-WMF and VIGNERON. It will be important to put a constraint on the property so that it is only used as a main statement that applies to the whole Lexeme. Switching to  Support. Deryck Chan (talk) 11:23, 20 June 2022 (UTC)

    I also have a translingual cautionary tale for you: Japanese pronouns form more of an open set than a closed set, and behave syntactically like regular nouns (Japanese verbs don't inflect for person or number). See Japanese pronouns and make sure your implementation can handle these. Deryck Chan (talk) 12:00, 20 June 2022 (UTC)

    Actually, switch to  Neutral as it's becoming clear that a conceptually consistent definition of this property will rely on grammatical concepts that don't exist in the languages I speak. Deryck Chan (talk) 13:44, 14 July 2022 (UTC)

    Thanks, @Deryck Chan!

    Insofar there is no verbal agreement in Japanese, one may say that there is no grammatical (morpho-syntactic) person feature at all in Japanese (in contrast to a semantic person category which does exist). See the section The status of 'person' as a feature in the Surrey Morphology Group page for a discussion of this. So one option would be not to mark this property at all on Japanese pronouns, and instead just use the sense linking. However, that may cause some confusion, so one could still mark the relevant pronouns as having the grammatical person feature, with the understanding that this is only a semantic feature in Japanese (and similar languages). AGutman-WMF (talk) 14:42, 20 June 2022 (UTC)

    In that case you need to think carefully whether this property is intended to tag semantics or alignment agreement. I think it would be more intuitive to tag semantics, because tagging alignment can get messy quite quickly (think of all the Latin-influenced languages where the honorific second person aligns syntactically with the third person). Deryck Chan (talk) 11:01, 23 June 2022 (UTC)

    @Deryck Chan I'm not sure what you mean here by Alignment, but in languages where we have person-agreement morphology, the grammatical feature should be used to mark that and not the possibly different semantic meaning. For instance, Spanish usted (L56997) should be marked as third-person (as it is now on the forms). To indicate the semantics, its sense could be linked to second person (Q51929049) (or even better, a "second-person formal" item, but this doesn't seem to exist). AGutman-WMF (talk) 08:27, 29 June 2022 (UTC)

    @AGutman-WMF: I mean person-agreement morphology indeed. But if we're tagging agreement morphology rather than semantics and doubling down on the classification of usted (L56997) as third person, we're back to my previous point: some languages like Chinese and Japanese have pronouns that have a clear semantic person, but no person-agreement morphology whatsoever in other parts of speech. User:VIGNERON - What do you think? Deryck Chan (talk) 13:50, 14 July 2022 (UTC)

    @Deryck Chan In languages where there is no person agreement morphology, there is no need to use this feature at all, as the semantic person should in principle be marked in the Senses section. AGutman-WMF (talk) 09:27, 18 July 2022 (UTC)

•  Weak support: is there anything wrong with Sense → item for this sense (P5137) → third-person singular (Q51929447)? would this property replace such a statement? -Shisma (talk) 16:51, 1 July 2022 (UTC)

  @Shisma Sense provides the semantics of a lexeme, but this is distinct from the grammatical feature. As @Deryck Chan pointed out above, in some languages, for instance, there are formal pronouns whose grammatical person feature is a third person (as apparent by verbal agreement) while the semantic person (the sense) is actually a second person. In general, I think that the semantics (senses) should be modeled differently than the grammatical features. AGutman-WMF (talk) 10:42, 4 July 2022 (UTC)

@AGutman-WMF, VIGNERON, Deryck Chan, Shisma:  Done --Fralambert (talk) 03:40, 13 September 2022 (UTC)

Thanks for the ping. I'll let AGutman and Vigneron sort out the debate on semantic person vs alignment person before I start using it. Deryck Chan (talk) 19:27, 21 September 2022 (UTC)