[11:46:30] How should we call for lexemes when rendering in language variants? For example, this query (https [11:46:30] //www.wikifunctions.org/view/en/Z6830?call=%7B%22Z1K1%22%3A%22Z7%22%2C%22Z7K1%22%3A%22Z6830%22%2C%22Z6830K1%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q11464%22%7D%2C%22Z6830K2%22%3A%7B%22Z1K1%22%3A%22Z6092%22%2C%22Z6092K1%22%3A%22P5137%22%7D%2C%22Z6830K3%22%3A%22Z1689%22 [11:46:30] %7D) returns nothing. Should we call for the variant then fallback to the parent language? [11:53:28] 🤔 Well, let’s think about how “pavement” vs “sidewalk” behaves. I don’t have an answer at my mental fingertips. (re @u99of9: How should we call for lexemes when rendering in language variants? For example, this query returns nothing. Should we call for ...) [11:59:01] What do you mean by variant? Synonyms or lexical forms or others? (re @u99of9: How should we call for lexemes when rendering in language variants? For example, this query returns nothing. Should we call for ...) [12:04:49] It may not be a good general term. I would say the variants of English include en-au, en-gb, en-us, en-simple, etc (re @higa4: What do you mean by variant? Synonyms or lexical forms or others?) [12:05:32] ah, I see. [12:06:30] The usual task is to get the right spelling of the right lexeme for a QID. But first we have to get to the right lexeme. (re @higa4: What do you mean by variant? Synonyms or lexical forms or others?) [12:08:21] L325162 isn't linked to an item. (re @Al: 🤔 Well, let’s think about how “pavement” vs “sidewalk” behaves. I don’t have an answer at my mental fingertips.) [12:09:56] L320843 is linked to *Q3352369* [12:11:03] L297541 is linked to Q177749 [12:12:20] All of them are just "en". So the outcome would depend on the QID alone... not great [12:12:27] Isn’t it the value of P5137? [12:12:51] Yes, that's the property used when I say "is linked to" (re @higa4: Isn’t it the value of P5137?) [12:13:58] It appears to be lacking the “sidewalk” sense. I don’t know what Australians call them! But, yeah, these are “en” lexemes; are there any exceptions? I’m inclined to doubt it. (re @u99of9: L325162 isn't linked to an item.) [12:14:19] Many lexemes don’t have this property yet. That’ my usual task in Japanese to add;) [12:14:26] we call them footpath! (re @Al: It appears to be lacking the “sidewalk” sense. I don’t know what Australians call them! But, yeah, these are “en” lexemes; are t...) [12:15:21] How very sensible! (re @u99of9: we call them footpath!) [12:16:03] Agreed. It will be great to have them linked. Most functions I write at the moment just use the item labels because lexemes are so rarely linked. But eventually we will check lexemes first, then default to labels. (re @higa4: Many lexemes don’t have this property yet. That’ my usual task in Japanese to add;)) [12:22:12] So it looks like it’s a case of item for this sense in •lexeme• language filtered by consideration of usage by language variant. In the case where the same lexeme is appropriate across variants, we’re filtering form-representations. I’m using “filter” quite loosely here, sorry. [12:28:30] L18155 has spelling variants in the lemma, like colour/color (from memory). (re @Al: So it looks like it’s a case of item for this sense in •lexeme• language filtered by consideration of usage by language variant....) [12:36:19] Yes, and the overall lexeme has both en and en-us. But the sense only has a gloss in en. Is that why the en-us query (https [12:36:19] //www.wikifunctions.org/view/en/Z6830?call=%7B%22Z1K1%22%3A%22Z7%22%2C%22Z7K1%22%3A%22Z6830%22%2C%22Z6830K1%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q11464%22%7D%2C%22Z6830K2%22%3A%7B%22Z1K1%22%3A%22Z6092%22%2C%22Z6092K1%22%3A%22P5137%22%7D%2C%2 [12:36:20] 2Z6830K3%22%3A%22Z1689%22%7D) doesn't find it? (re @Al: L18155 has spelling variants in the lemma, like colour/color (from memory).) [12:37:32] Oh, no, maybe the overall lexeme is just in en: : https://tools-static.wmflabs.org/bridgebot/fa7ff0a6/file_81158.jpg [12:38:00] I think it is just because the *lexeme* language is “en”. (re @u99of9: Yes, and the overall lexeme has both en and en-us. But the sense only has a gloss in en. Is that why the en-us query doesn't fin...) [12:49:51] Okay. I've added L320843-S2 and L325162-S2. I'll leave the functions for now. Anyone should feel free to take them on. [13:07:25] Is there a way to indicate on Wikidata that a sense applies only to a particular language variety? [13:35:17] It’s more geographical, I think: P6084 (re @dvd_ccc27919: Is there a way to indicate on Wikidata that a sense applies only to a particular language variety?) [13:37:05] But also P7481. Who knew? (re @dvd_ccc27919: Is there a way to indicate on Wikidata that a sense applies only to a particular language variety?) [13:40:43] Grouped under https://www.wikidata.org/wiki/Special:WhatLinksHere?target=Q116547761&namespace=120&limit=50 (re @Al: But also P7481. Who knew?) [13:44:05] I've added this to the senses. Note that they link to QIDs Q7976 Q44679 and Q7979. So we'll need to do additional work to map from the ZIDs or language codes. (re @Al: But also P7481. Who knew?) [13:46:59] Ah, well done… *T344170* (re @u99of9: I've added this to the senses. Note that they link to QIDs Q7976 Q44679 and Q7979. So we'll need to do additional work to map f...) [14:06:29] Thanks, I've commented there. Low priority! (re @Al: Ah, well done… T344170) [14:53:00] 7380 [15:36:23] Z32599 should work in these cases (re @u99of9: How should we call for lexemes when rendering in language variants? For example, this query returns nothing. Should we call for ...) [15:36:53] Now I've modified Z27327 to also handle the footpath/pavement situation [16:15:07] There is Z29649 that can do the conversion. It is still dependant from a hardcoded list, which is not ideal (re @u99of9: I've added this to the senses. Note that they link to QIDs Q7976 Q44679 and Q7979. So we'll need to do additional work to map f...)