Wikidata:Property proposal/cognate

cognate edit

Originally proposed at Wikidata:Property proposal/Lexemes

Not done

Description	Lexeme in another language derived from the same word in an old language
Represents	cognate (Q690548)
Data type	Lexeme
Domain	Lexeme
Example 1	hound (L6419) → hund (L31499)
Example 2	bryn (L252176) → brow (L16178)
Example 3	father (L3624) → père (L2245)
See also	derived from lexeme (P5191)

Begrundelse edit

Etymological information is conveyed by the derived from lexeme (P5191) property. Parallel to this we could establish a property, the proposed property, where two (modern) languages could indicate mutual derivational root. This has the NxN downside that is also an issue with the translation (P5972) property, where all cognates in other languages need to be specified on all lexemes. However, in some cases it may be easier to establish cognates rather than the complete derivation chain to find cognates via SPARQL. Finn Årup Nielsen (fnielsen) (talk) 00:52, 28 February 2020 (UTC)[reply]

Discussion edit

Support I think this is reasonable as long as people don't feel a need to fill in all possible cognates... ArthurPSmith (talk) 19:19, 28 February 2020 (UTC)[reply]
Weak oppose I think this is something that can be obtained using SPARQL queries. Also, who will decide which cognates should be included in the new property? --Tinker Bell ★ ♥ 23:07, 28 February 2020 (UTC)[reply]
- Yes, the cognate can be obtained with SPARQL queries, - if the data is set up. However, some source such as The Concise Dictionary of English Etymology (Q58489371) specifies derived words for English, but cognate (Q690548) for other Germanic (Q21200). From a source point of view, it would be easier with a cognate property. The guess the number of cognate could potentially be large for certain Indo-European words (arm?, mother?) and possible modern English loadwords, but I suppose that is the same problem with the translation property. — Finn Årup Nielsen (fnielsen) (talk) 13:48, 4 March 2020 (UTC)[reply]
Weak oppose per Tinker Bell, plus the number of cognate can be very large, like 10k large and it's not exceptional. To take the first example, we already have lexeme for hound (L6419), hundo (L31197), hund (L31499), hund (L32379), Hund (L34727), Hund (L184269), hund (L5670), hond (L208466), hond (L208467), hont (L220837), hunt (L220841), *hundaz (L220847), hunt (L220848), *kun-tós (L220849), *ḱwn̥tós (L220850) (and probably much more I just did a quick search, with 15 lexemes it's already 105 links, n(n-1)/2 as its a complete graph ; for 100 cognates it's 4950 links). Cdlt, VIGNERON (talk) 18:12, 30 March 2020 (UTC)[reply]
Support cognates can help building etimologies, and explaining how a word is related to another. -Theklan (talk) 13:02, 22 May 2020 (UTC)[reply]
Oppose Per above. You could literally have 100 values for this property. --IWI (talk) 23:42, 13 September 2020 (UTC)[reply]
@Fnielsen, Tinker Bell, VIGNERON, Theklan, ImprovedWikiImprovment: Not done Stale and no consensus on creation. If proposed again, please reference this discussion. ArthurPSmith (talk) 16:46, 11 January 2021 (UTC)[reply]