Wikidata:Property proposal/feminine form

feminine form of lexeme

edit

Originally proposed at Wikidata:Property proposal/Lexemes

   Not done
Descriptionlexeme that is the feminine form of the subject lexeme
Representsfeminine (Q1775415)
Data typeLexeme
Domainlexeme
Example 1king (L9670)queen (L1380)
Example 2acteur (L13374)actrice (L12849)
Example 3padre (L221496)madre (L47362)
Example 4lion (L17815)lioness (L43104)
Planned useto be used where it doesn't make sense to add the feminine form as a Form to the subject Lexeme, but instead is considered to be a separate Lexeme
See alsofemale form of label (P2521), male form of label (P3321)

Motivation

edit

Currently there is no single way to do this. Theklan (talk) 13:47, 25 October 2019 (UTC)[reply]

Discussion

edit

  Comment I think the general idea makes sense, but it shouldn't be limited to feminine / masculine form (many languages have more than these two genders, and some don't have both of these but have others). A more generic property "gendered version" or whatnot might make sense :) (the word "form" might not be ideal because we already use it for forms of a lexeme).--Reosarevok (talk) 15:38, 25 October 2019 (UTC)[reply]

  Comment @Theklan: This relationship can already be represented in the Item space by using female form of label (P2521) and male form of label (P3321), and at the Sense level by using antonym (P5974) qualified with criterion used (P1013) = gender binary (Q5530970). It could be a good to have similar properties for the Lexeme level. Perhaps they should be called "female form of lexeme" and "male form of lexeme" to better mirror the existing Item-typed properties, and to specify that it is at the Lexeme level rather than the Form level (as the current proposed name is a bit ambiguous in that regard). Also, it would be good to include a few more use cases other than just the "king & queen" example. Liamjamesperritt (talk)

Maybe the solution is to have a general gender representation property, so we can have more than two options. Also I don't consider that actor and actress are antonyms. -Theklan (talk) 08:23, 29 October 2019 (UTC)[reply]

  Comment @Theklan, Reosarevok, Fnielsen, ArthurPSmith: With permission from Theklan, I updated the proposal to make it a little clearer. Liamjamesperritt (talk) 05:18, 9 November 2019 (UTC)[reply]

The symmetric proposal is here: Wikidata:Property proposal/masculine formFinn Årup Nielsen (fnielsen) (talk) 12:38, 13 November 2019 (UTC)[reply]

  Comment This property should be used on sense level not lexeme. Dichotomy might not apply to all senses of a given lexeme. --Lexicolover (talk) 22:04, 4 December 2019 (UTC)[reply]

  Comment The proposed name for the property is not correct. King/queen are not forms of the lexeme, these are two separate lexemes describing close concepts which differ by gender of the subject. Thus something like "male/female describing term for this sense" should be better. --Lexicolover (talk) 22:04, 4 December 2019 (UTC)[reply]

  Comment Is it possible to generalize this concept to cover cases like sheep/ram/lamb etc.? --Lexicolover (talk) 22:04, 4 December 2019 (UTC)[reply]

  Comment All your proposals are really interesting. We need something to denote this (sheep/ram/lamb is a good example) and antonym is not valid (queen is not the antonym of king). Can someone make a proposal that could handle everything? -Theklan (talk) 15:11, 18 December 2019 (UTC)[reply]

  Strong oppose The proposal conflates, as it seems, multiple aspects that are customarily discussed as separate phenomena, and usefully so in both linguistics and its practical applications; and needlessly inscribes a concrete notion of ‘gender’ that applies only to a few languages.

  • king (L9670) vs. queen (L1380) is a question of mere semantics and so should operate on the level of senses. This could be described using item for this sense (P5137); as an example, see Gräfin (L34190). This could then be queried like so:
    SELECT ?l1 ?lemma1 ?sense1 ?gloss1 ?l2 ?lemma2 ?sense2 ?gloss2 WHERE {
      # Assuming that we have a given sense ?sense1 (here: L34190-S1),
      FILTER (?sense1 = wd:L34190-S1)
    
      # we are looking for a ?sense2 that has the same conceptualized meaning ?meaning,
      ?sense1 wdt:P5137 ?meaning.
      ?sense2 wdt:P5137 ?meaning.
      
      # but we don’t want the ?meaning to be the gender aspect of the sense.
      MINUS { 
        ?meaning wdt:P31 wd:Q48277, wd:Q48264
      }
    
      # We also want the ?sense2 to be about a lexeme in the same language
      ?l1 dct:language/^dct:language ?l2.
      # where ?l1 and ?l2 are the lexemes for the senses ?sense1 and ?sense2, respectively  
      ?sense1 ^ontolex:sense ?l1.
      ?sense2 ^ontolex:sense ?l2.
      # (Note that we would write the above three triples as the single line
      ?sense1 ^ontolex:sense/dct:language/^dct:language/ontolex:sense ?sense2.
      # but then we wouldn’t be binding lexemes ?l1 and ?l2, which we might want for debugging.)
    
      # And we don’t want ?sense2 to be identical to the original ?sense1
      FILTER (?sense2 != ?sense1)
                                  
      # Finally, for debugging purposes, we’re binding the lemmas and glosses for ?sense1 and ?sense2
      ?l1 wikibase:lemma ?lemma1.
      ?sense1 skos:definition ?gloss1.
      ?l2 wikibase:lemma ?lemma2.
      ?sense2 skos:definition ?gloss2.
    }
    LIMIT 1
    
    Try it!
    (Technically, while the main code is agnostic to specific genders, the above is a query that retrieves a ‘masculine form’ because it is applied to a ‘feminine form’ and the only other ‘form’ currently defined is one that may be viewed as either masculine or generic. But the query would work in reverse as well, although the lexeme properties might not be fully populated yet.) (Incidentally, @Lexicolover: This could also address the case of “sheep” vs. “ram” vs. “lamb”; in the case of the latter in particular by substituting baby animal (Q12038335) for the gender items.)
  • acteur (L13374) vs. actrice (L12849) might be about etymology, and so in regard to that should operate on the level of lexemes using derived from lexeme (P5191): actrice (L12849)derived from lexeme (P5191)acteur (L13374), maybe with a qualifier like mode of derivation (P5886)-trice (L25480) or using some better-suited property (maybe one more closely related to gender inflection (Q1124523), whose current label “gender inflection” is maybe not quite on point):
    actrice (L12849)derived from lexeme (P5191)acteur (L13374)mode of derivation (P5886)-trice (L25480)
    lioness (L43104)derived from lexeme (P5191)lion (L17815)mode of derivation (P5886)
    -ess

    A query equivalent to what this proposed property might have been intended to be about in regards to etymology might search for, depending on the notion of ‘gender’ applicable to the language in question,
  • The general nonapplicability of a dichotomous notion of ‘gender’ has already been pointed out. I might add that even under the assumption of a dichotomous (real-world) gender,
  • @Theklan: If indeed a “single way to do this” is desired, one might build a query (and maybe service around that) that somehow combines the two queries sketched above, though it is not clear to me how that would yield a well-defined result in the case of languages that, for example, have both (real-world-)gender-dependent lexemes and (language-intrinsic) grammatical gender.

BlaueBlüte (talk) 07:38, 26 December 2019 (UTC)[reply]