Wikidata:Property proposal/lexeme of other gender

lexeme for other gender edit

Originally proposed at Wikidata:Property proposal/Lexemes

Not done

Description	identifier for a lexeme of another grammatical gender
Data type	Lexeme
Domain	lexeme
Example 1	Parisien (L26359) → Parisienne (L25620)
Example 2	Parisienne (L25620) → Parisien (L26359)
Example 3	Arzt (L21064) → Arzt (L21064)
Example 4	Ärztin (L21065) → Ärztin (L21065)
Example 5	traductor (L494274) → traductora (L739178)
Example 6	traductora (L739178) → traductor (L494274)
Example 7	Française (L592251) → Français (L592250)
Example 8	Français (L592250) → Française (L592251)
Example 9	brasileira (L739183) → brasileiro (L477181)
Example 10	brasileiro (L477181) → brasileira (L739183)
Example 11	Brasilianer (L626636) → Brasilianerin (L626637)
Example 12	Brasilianerin (L626637) → Brasilianer (L626636)
Example 13	bonic (L739312) → bonica (L739314)
Example 14	bonica (L739314) → bonic (L739312)
Planned use	added to related lexemes when editing or creating them

Motivation edit

For languages for which separate lexemes have been created for different grammatical genders, there is no property to link the lexemes. Sometimes the property combines lexemes (P5238) has been used, but in many cases there is no obvious link between related lexemes. This property would provide a way to explicitly link these lexemes. UWashPrincipalCataloger (talk) 19:39, 20 November 2022 (UTC)[reply]

Discussion edit

Comment I think most of these examples could be merged, especially the adjective ones. A property like this should probably link senses to each other rather than lexemes. Depending on the criteria used, senses for semantically male or female counterparts could be considered antonyms or synonyms. That way for example senses on single lexemes can be linked to each other like invalid ID (L702427#S2) to invalid ID (L702427#S3). Linking senses like this also avoids conflation of grammatical and semantic gender. Cats of unspecified gender or female cats are feminine بلیاں and male cats (tomcats) are masculine بلے. However, peacocks of unspecified gender or male peacocks are masculine مور and female peacocks (peahens) are feminine موریاں. There is no “rule” we can rely on to determine the grammatical gender required for the unspecified sense.

We do already have the property hyperonym (P6593) which can be used to link gendered senses to an unspecified one and semantic gender (P10339) with which we can link senses to gendered values. (It would then be possible to query for senses linked to the same item with differing semantic gender values.) A new property might be helpful though for cases where the opposite gender senses are not linked to the same item. -عُثمان (talk) 23:51, 21 November 2022 (UTC)[reply]

Comment agreed. this should link senses. – Shisma (talk) 16:05, 28 November 2022 (UTC)[reply]

Question @UWashPrincipalCataloger: what is the use-case for the ‘obvious link’ between the lexemes of various grammatical genders? How this problem is best solved depends to a large extent on how and where this kind of link is to be used. ―BlaueBlüte (talk) 14:05, 30 January 2023 (UTC)[reply]
I think I was just wanting to be able to relate two or more lexemes that are only different based on the gender implied. Perhaps I should have said "there is no explicit link between related lexemes" instead of using the word "obvious". Ultimately, I'd like to know what the best practice is for relating traductor to traductora or Brasilianer to Brasilianerin, etc. If there is already a recommended way to accomplish this, is it documented somewhere? AdamSeattle (talk) 04:45, 31 January 2023 (UTC)[reply]
Comment This proposal seems unclear as to what kind of ‘gender’ it refers to. It states “grammatical gender”, but the examples are all about male/female ‘natural gender’ in cases where there happens to be a predictable correlation between ‘natural gender’ and grammatical gender. See also comment above by @عُثمان:. Also, all examples are such that the (pairs of) lexemes are clearly derived from one another, thus missing cases like “king” vs. “queen”, “ewe” vs. “ram”. ―BlaueBlüte (talk) 14:05, 30 January 2023 (UTC)[reply]
Comment The proposal as stated creates a many-to-many relationship among lexemes (or senses, if modified as suggested above). This should be avoided in favor of one-to-many or one-to-one relationships. (Even in languages with a fairly low number of (grammatical or ‘natural’) genders there are often still multiple lexemes per gender for parallel forms derived in different ways, with different suffixes, etc., all of which would have to linked to all such forms for all other genders.) ―BlaueBlüte (talk) 14:05, 30 January 2023 (UTC)[reply]
Comment A while ago a similar property but specifically for “feminine” gender was proposed but not accepted. This new proposal is an improvement over the earlier proposal in that it does not inscribe into Wikidata properties a specific set of genders (as did the earlier proposal with something like “feminine” vs. (implied) ‘other’ gender). But even this new proposal might still benefit from looking into ways to link senses rather than lexemes, for which I suggested a method in the discussion of the earlier proposal. (That method suggested recording senses’ gender connotation using item for this sense (P5137). Since then, the creation of semantic gender (P10339) has opened up another possibility.) Note that that method would not (only) link gender-connotated senses via gender-unspecific ones (as suggested by عُثمان above) but via items indicated via item for this sense (P5137). Techniques for querying are also detailed here. This is where knowing the use-case would be helpful in assessing whether such queries would be practical. ―BlaueBlüte (talk) 14:05, 30 January 2023 (UTC)[reply]
Comment Note also an old informal proposal regarding ‘gendered forms of labels’ highlighting some issues that might be worth considering here as well. ―BlaueBlüte (talk) 14:11, 30 January 2023 (UTC)[reply]

@UWashPrincipalCataloger, عُثمان, BlaueBlüte: Not done there is clearly no consensus. As pointed in the comments, the topic is complex (grammatical gender, natural/semantic gender, social gender, etc.) and conversely there is other property that can fill most need. Cheers, VIGNERON (talk) 10:05, 29 April 2023 (UTC)[reply]