Wikidata:Property proposal/langue mère

parent language edit

Originally proposed at Wikidata:Property proposal/Generic

Not done

Description	languages from which this language derive, may be a proto-language. Examples are Proto-Germanic (PGmc) or Germanic parent language (GPL).
Represents	parent language (Q94583221)
Data type	Item
Domain	linguistics (Q8162)
Allowed values	valueinstance of (P31)languoid (Q17376908)
Example 1	French (Q150) → Middle French (Q1473289)
Example 2	English (Q1860) → Early Modern English (Q1472196)
Example 3	Ancient Greek (Q35497) → Mycenaean Greek (Q668366)
Example 4	Afrikaans (Q14196) → Dutch (Q7411)
Source	référence externe, article de liste de Wikipédia.
Planned use	Sur les langues d’Oïl et indo-européennes dans un premier cas, mais servira vite pour toutes les langues
See also	Wikidata:Property proposal/langue fille langue sœur, based on (P144), derivative work (P4969)

Motivation edit

Je suis en train de classifier les langues et notamment leurs relations diachroniques entre elles. Cette propriété sera vraiment utile pour lier les langues et générer des arbres linguistiques en diachronie. Lyokoï (talk) 22:47, 16 May 2020 (UTC)[reply]

Notified participants of WikiProject Linguistics

Discussion edit

~~Weak support~~ currently this role is being served by subclass of (P279). I think the case can be made that this is better. Iwan.Aucamp (talk) 10:14, 17 May 2020 (UTC)[reply]

~~Strong support~~ Upgrading to strong support. I'm pretty sure subclass of (P279) is not ideal for modelling this. Iwan.Aucamp (talk) 21:21, 17 May 2020 (UTC)[reply]

Weak oppose Revised my support. I favour using languagebased on (P144)parent languagecriterion used (P1013)parent language (Q94583221) Iwan.Aucamp (talk) 18:44, 19 June 2020 (UTC)[reply]

@Iwan.Aucamp: There is a big problem for me for (P144), it’s for construct things, not natural thing. This isn’t a conscient choice for anturals languages to have one or more parent language. (P144) is for human work (for artificial language for example), not for natural thing. I didn’t know how futur french are based on, maybe on actual french, but we can’t say more ! But for my new artwork, I always know on what I had based my work. It’s an important differency for me. Lyokoï (talk) 19:03, 19 June 2020 (UTC)[reply]

Weak support for this property, Oppose for this name as it is easily confused with native language (P103).--GZWDer (talk) 10:19, 17 May 2020 (UTC)[reply]
- @Lyokoï, GZWDer: How about "parent language" for the English label? Another possibility is "proto-language" but I think "parent language" may be more appropriate - I think "proto-language" is a kind of "parent language". Iwan.Aucamp (talk) 20:48, 17 May 2020 (UTC)[reply]
  - @Iwan.Aucamp, GZWDer: I’m okay for « parent language », it is better I think, no problem for this change. But « proto-language » isn’t the same notion, because proto-language are only reconstructed language. And yes, « proto-language » is a kind of » parent language ». Lyokoï (talk) 20:57, 17 May 2020 (UTC)[reply]
Comment @Lyokoï: I think the current value for 'subject item', which is parent language (Q94583221), is not right. If you look at other properties, this should be the item that values for this property should be instances of. For example, for writing system (P282) it is writing system (Q8192) so all values for writing system (P282) must have ???instance of (P31)writing system (Q8192). The way it is now all values for this property will have to be ???instance of (P31)parent language (Q94583221) and I don't think that is entirely ideal. I think it would be better to make 'subject item' languoid (Q17376908). Iwan.Aucamp (talk) 21:14, 17 May 2020 (UTC)[reply]
- Actually no, my apologies, it is right as is. I think allowed values need some work though, will do it shortly. Iwan.Aucamp (talk) 21:18, 17 May 2020 (UTC)[reply]
  - @Iwan.Aucamp: Yep, thank you. Lyokoï (talk) 08:02, 18 May 2020 (UTC)[reply]
Comment @Lyokoï: I removed 'expected completeness' (expected completeness (P2429)), this should only be used for identifiers I believe (see Property_talk:P2429#Should_only_be_used_on_identifiers). Iwan.Aucamp (talk) 21:40, 17 May 2020 (UTC)[reply]
- @Iwan.Aucamp: Okay. No problem for me. Lyokoï (talk) 08:02, 18 May 2020 (UTC)[reply]
Comment @Lyokoï: I looked at Afrikaans (Q14196), and it has Afrikaans (Q14196)based on (P144)Dutch (Q7411), would this not be sufficient for your needs? See this query:
```
#defaultView:Graph
SELECT ?item ?itemLabel ?parentLanguage ?parentLanguageLabel WHERE {
  ?item wdt:P31 / wdt:P279* wd:Q17376908.
  ?item wdt:P144 ?parentLanguage.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
```
Try it!
Iwan.Aucamp (talk) 21:50, 17 May 2020 (UTC)[reply]
@Iwan.Aucamp: Hello, the property P144 isn’t good for lots of languages. Afrikaans is a pidgin due to the mix of some languages. But French isn’t « based of » middle french, it is directly from this parent language. I see P144 like a act due to choices, but most of languages are natural and not based of choice for their evolution. For me Afrikaans have some parent language. Lyokoï (talk) 08:02, 18 May 2020 (UTC)[reply]
I would add that according to the description P144 seems to relate only to works, only due to choices therefore, which is not the case with natural languages. Lyokoï (talk) 08:08, 18 May 2020 (UTC)[reply]
Oppose I think using based on (P144) is sufficient. We use it in computer languages to express the same relation. --Tinker Bell ★ ♥ 02:37, 18 May 2020 (UTC)[reply]
- @Tinker Bell: Did you have references about that, I’m curious ? Lyokoï (talk) 08:02, 18 May 2020 (UTC)[reply]
  - @Lyokoï: I generally agree with Tinker Bell. I get that the word "based on" may not be that great - but it can capture the relation that you are looking for I believe, we can maybe expand based on (P144) to accommodate the case better but I think an important principle is to reduce the proliferation of properties as much as possible in favour of using more general properties and qualifiers where it is not enough. If there are different "based on" like relations you need to capture, maybe that is a good basis for a new property, but I guess you can put multiple languages in the property if you need to, and for cases where there is more than one kind of "based on" relation qualifiers like criterion used (P1013) or determination method (P459) may work. For example Afrikaans (Q14196)based on (P144)Dutch (Q7411)criterion used (P1013)parent language (Q94583221) or Afrikaans (Q14196)based on (P144)Dutch (Q7411)determination method (P459)parent language (Q94583221). Here is a query to illustrate how it works for programming languages
    #defaultView:Graph SELECT ?item ?itemLabel ?parentLanguage ?parentLanguageLabel WHERE { ?item wdt:P31 / wdt:P279* wd:Q9143. ?item wdt:P144 ?parentLanguage. SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". } }
    Try it!
    Iwan.Aucamp (talk) 15:29, 18 May 2020 (UTC)[reply]
    - I understand that, but I think it is not a good way for this property. As I say before, the parental relation between two natural languages is determined by lots of studies because it is a very complex notion. This link, can be resume by a link between to work like P144. The parental relation need its own qualifiers like « concerned linguistic movement » or « based on lexical studies, or historical pronunciation studies, etc. » (Sorry, I didn’t use the WD language here, I just try to explain the difference). The fact that language are naturals and works not should be treated differently. Because the target objects are fuzzy, with imprecise limits, constantly evolving and always studied. The objects concerned by P144 are precise, chosen and of a much smaller range. Finally, and the most important argument I think : the relationship that I propose carries an exclusive value of diachrony. The parent languages must have appeared before the daughter languages. This notion is absent from P144. --Lyokoï (talk) 19:02, 18 May 2020 (UTC)[reply]
Comment The WikiProject Languages has not yet been notified. Visite fortuitement prolongée (talk) 19:35, 18 May 2020 (UTC)[reply]
Comment Currently this is managed by based on (P144) (such Tsotsitaal (Q1472421) P144 Afrikaans (Q14196) P144 Dutch (Q7411)) and follows (P155) (such Modern Greek (Q36510) P155 Medieval Greek (Q36387) P155 Koine Greek (Q107358) P155 Ancient Greek (Q35497) P155 Mycenaean Greek (Q668366) P155 Proto-Greek (Q1231805)). Visite fortuitement prolongée (talk) 19:35, 18 May 2020 (UTC)[reply]
Support A parent language is a much more specific relationship than based on (P144) since it does not just imply "borrowing" some elements from a langage but inheriting core morphological traits (that's the main reason that phylogenetic trees are usable in historical linguistic). Since we have many items on languages and dialects I believe the new property would fit a real need. Alexander Doria (talk) 15:43, 19 May 2020 (UTC)[reply]
Support As proposer. Lyokoï (talk) 16:29, 19 June 2020 (UTC)[reply]
Weak support, good idea but as the notion is very specific, we need good constraint and checking to make sure the property is correctly used. parent language (Q94583221) could use some improvements too (thanks already @Iwan.Aucamp: ). Cheers, VIGNERON (talk) 17:33, 19 June 2020 (UTC)[reply]
@Lyokoï, Iwan.Aucamp, GZWDer, Tinker Bell: @Visite fortuitement prolongée, Alexander Doria, VIGNERON: Not done Stale proposal; however there was some expression of support. Please reference this discussion if you make a new proposal along these lines. ArthurPSmith (talk) 17:02, 11 January 2021 (UTC)[reply]
@ArthurPSmith: Okai, what did this property need to be accepted ? Lyokoï (talk) 18:47, 11 January 2021 (UTC)[reply]
@Lyokoï: Clearer expressions of support would have helped - that might have happened if you had been a bit clearer about why based on (P144) was insufficient from the start. You might also try to find other people interested in this (such as those who commented) and work on crafting a better proposal with them. ArthurPSmith (talk) 19:37, 11 January 2021 (UTC)[reply]
@ArthurPSmith: I will take the time to explain with a better proposal and with people interested. Thanks ! Lyokoï (talk) 19:47, 11 January 2021 (UTC)[reply]