Wiktionary:Votes/2023-11/Ordering of descendants in mainspace entries

Voting on: adding the following text to WT:Descendants regarding the ordering of descendant terms in mainspace entries:

Inherited terms should be listed first, ordered according to any layout specified on the relevant language considerations page (so far as one is specified), and otherwise alphabetically by language name or family at each indentation level. These are then followed by all other descendants (e.g. borrowed terms), alphabetically by language name at each indentation level. For languages with multi-word names, their place in the order is determined by the first letter of the first name component – for instance, Norwegian comes before Old English, which comes before Portuguese.

Note: (outdated; see next note) this vote is about mainspace entries. Reconstruction entries, as far as I can tell, can and do have their own conventions for organizing descendants.

Note #2: a change has been proposed to the text to better reflect language-specific considerations, and the handling of descendants of descendants. That also subsumes previous concerns about reconstruction entries. I've incorporated the change into the wording of this proposal.

Right now, descendant terms in mainspace entries are ordered in various ways, at no easily discernible benefit to editors or users:

• in some entries, descendants are ordered as per this vote's proposal, e.g. Ottoman Turkish قادر (kadir, “able, powerful”).
• in other entries, all descendants (inherited, borrowed, etc) are ordered alphabetically relative to one another, e.g. Latin scorpiō (“scorpion”).
• for languages with multi-word names, the sort key is variously the first letter of the first word, or some other letter, e.g. Persian بازار (bâzâr, “marketplace”) vs Arabic عَقْل (ʕaql, “reason; mind”).

The goal of this vote is to confirm that the proposal captures a community best practice (as has been indicated on BP), and if so - communicate it via WT:EL.

Schedule:

• Vote starts: 00:00, 5 December 2023 (UTC)
• Vote ends: 23:59, 3 January 2024 (UTC)
• Vote created: Chernorizets (talk) 02:45, 28 November 2023 (UTC)

Discussion:

• Wiktionary:Beer parlour/2023/October § Ordering of descendants within a "Descendants" section in an entry
• Wiktionary talk:Votes/2023-11/Ordering of descendants in mainspace entries

1. Support - logical, and mostly what we do already. Theknightwho (talk) 15:19, 5 December 2023 (UTC)
2. Support --Vahag (talk) 18:49, 5 December 2023 (UTC)
3. Support – presuming that “derived” is not interpreted in an exaggerated, merely formal fashion: some “derived” terms in children languages may be felt more inherited. I disagree with the 3rd sentence, following the “main component of language name” practice, which practice is to a large part also my fault, but at least it makes the order easily bottable, so there is less to think and argue. Fay Freak (talk) 00:50, 6 December 2023 (UTC)

   @Fay Freak that was precisely my intent with the proposed way of handling multi-word names - easily bottable, less to bicker about, less editorial judgment required. Multi-word names include things like Sranan Tongo, which is not a variety of "Tongo", and Lule Sami, where "Sami" is a language family name, while Lule is the meaningfully distinguishing part of the language. With 4,300+ languages on EN Wiktionary, I don't expect every editor to know every such detail, so I thought simpler would be better. Chernorizets (talk) 01:31, 6 December 2023 (UTC)

   @Chernorizets: One can formulate a more precise exception later, e.g. qualifiers in language names referring to age and geographic direction should be discounted. Middle Armenian with Armenian, Western Mari and Eastern Mari at Mari, Old South Arabian and Ancient North Arabian with Arabic. Fay Freak (talk) 02:01, 6 December 2023 (UTC)

4. Support as proposer. Chernorizets (talk) 06:12, 6 December 2023 (UTC)
5. Support: Enshrines existing practices. AG202 (talk) 15:16, 6 December 2023 (UTC)
6. Support: It's not really changing anything. The oppose and abstain votes all seem to be based on loopholes with the wording and unintended interpretations. Surely those can be fixed if they're really an issue. MedK1 (talk) 15:25, 7 December 2023 (UTC)
7. Support. Whoop whoop pull up ^{Bitching Betty ⚧️ Averted crashes} 21:09, 7 December 2023 (UTC)
8. Support — Fenakhay ^{(حيطي · مساهماتي)} 19:20, 8 December 2023 (UTC)
9. Support Vininn126 (talk) 20:07, 8 December 2023 (UTC)
10. Support The new text satisfies my concerns. -- King of ♥ ♦ ♣ ♠ 21:53, 8 December 2023 (UTC)
11. Support with new text. Thanks to Chernorizets for willingness to workshop an improvement. This, that and the other (talk) 09:29, 9 December 2023 (UTC)
12. Support. —Al-Muqanna المقنع (talk) 10:44, 9 December 2023 (UTC)
13. Support. I thought this was already policy. Kiril kovachev (talk・contribs) 17:57, 11 December 2023 (UTC)
14. Support — Vorziblix (talk · contribs) 15:21, 13 December 2023 (UTC)
15. Support, now that @Victar's concern has been addressed. —Caoimhin ceallach (talk) 15:59, 14 December 2023 (UTC)
16. Support a little bit conflicted since this proposal would make "varitey-LANGUAGE" languages be listed by the variety instead of the language (Old English and Old Korean would both be listed as "O" instead of "E" and "K"). That being said I don't feel too strongly about that and exceptions to the rules could make things messy. So you could argue that standardizing the listings without exceptions is probably the best thing to do to avoid all that hassle. So I support this proposal despite my minor reservations. - سَمِیر | Sameer (^{مشارکت‌ها} · ^بحث) 09:19, 15 December 2023 (UTC)
17. Support I also thought "Old English" sorts by "E", as I've seen many entries like that, but I personally don't mind as long as there is consistency, which this proposal will enhance. tbm (talk) 04:45, 21 December 2023 (UTC)

1. Oppose Many language families have their own sort rules, like seen at WT:AGEM, and are often ordered by language family, not strictly by the language name. It should be up to each language to control this. --{{victar|talk}} 04:43, 6 December 2023 (UTC)

   @Victar I had given the caveat that this proposal is about mainspace articles, not reconstruction articles, but it wasn't reflected in the proposed text. @AG202 spotted that as well, and now I've made it explicit in the text. Chernorizets (talk) 05:39, 6 December 2023 (UTC)

   Oppose The proposed text draws a false distinction between Reconstruction entries and mainspace entries. This misses the real point. What matters is not whether the language in question is reconstructed or not - the real distinction is between languages with large numbers of descendants organised into (sub-)families (Latin is the best example of this) and those with smaller numbers of descendants (like Old French) or ad hoc borrowings (like English). If this vote passes as is, it would lead to the perverse scenario where the descendants of Vulgar Latin reconstructions can be grouped by subfamily but descendants of attested Latin words cannot. I support the principle but the drafting needs more thought - my suggestion would be make the sorting requirement only apply to descendant lists where the descendants are not organised by subfamily. This, that and the other (talk) 04:52, 7 December 2023 (UTC)

   @This, that and the other I've started a thread on the talk page based on my understanding of what you and RichardW57m (abstain vote below) are saying - feel free to chime in if you'd like. My proposal doesn't entail a "flattening" of descendant trees, but it may not be obvious from the current wording. Thanks! Chernorizets (talk) 05:05, 7 December 2023 (UTC)

   @Chernorizets I'm talking about things like pullus#Descendants. The text of the proposed rule, as written, is clear - it means we would have to remove the family labels (which are not "languages" per the text) and present all languages as a flat list. Moreover, in many entries like this, the subfamilies are ordered in a non-alphabetical manner (here, the ordering appears to be geographical - roughly counterclockwise).

   I can only support if the proviso about Reconstruction pages is removed and replaced with text like "The alphabetical order is enforced at each level of nesting in the descendant list, except where a nesting level consists of linguistic subfamilies which Wiktionary does not treat as distinct languages (e.g. Ibero-Romance, West Slavic). <new line> This rule does not apply if a specific ordering for descendants is set out on the relevant ]; see ] for an example." (The last bit is to address Victar's concern.) This, that and the other (talk) 22:54, 7 December 2023 (UTC)

   @Theknightwho @Vahagn Petrosyan @Fay Freak @Chernorizets @Whoop whoop pull up @AG202 @MedK1 would any of you oppose a change to the wording as stated above? This, that and the other (talk) 23:01, 7 December 2023 (UTC)

   @This, that and the other I would prefer a slightly simpler wording. Change:

   Inherited terms should be listed first, ordered alphabetically by language name.

   to

   Inherited terms should be listed first, ordered according to any layout specified on the relevant language considerations page (so far as one is specified), and otherwise alphabetically by language name or family at each indentation level.

   The problem with your current wording is that if there's a layout specified on the language considerations page, there's still no obligation to fill in the gaps using alphabetical order (since that rule is overridden in its entirety).

   The next sentence then needs to be changed slightly:

   These are then followed by all other descendants (e.g. borrowed terms), alphabetically by language name at each indentation level.

   Firstly, it removes the slightly odd wording about terms being borrowed, derived or calqued (I'm not sure why these are necessary to specify, and the list is implied to be complete); and it also removes the word "also" from "also alphabetically".

   Theknightwho (talk) 23:18, 7 December 2023 (UTC)

   @Theknightwho Thanks. This wording is definitely simpler, and just as effective in practice. It does remove my proposed proviso allowing subfamilies to be in non-alphabetical order even when this order is not prescribed on the AXX page, but the solution is to simply add a section to each AXX page listing the conventional order of subfamilies. This is actually advantageous because it imposes a consistent ordering of subfamilies across all entries.

   Do you also support removing the carve-out for Reconstruction pages? It seems redundant with this change. This, that and the other (talk) 23:24, 7 December 2023 (UTC)

   @This, that and the other Yes. Theknightwho (talk) 23:36, 7 December 2023 (UTC)

   For clarity for those I pinged above, the proposed new text workshopped here is:

   Inherited terms should be listed first, ordered according to any layout specified on the relevant language considerations page (so far as one is specified), and otherwise alphabetically by language name or family at each indentation level. These are then followed by all other descendants (e.g. borrowed terms), alphabetically by language name at each indentation level. For languages with multi-word names, their place in the order is determined by the first letter of the first name component – for instance, Norwegian comes before Old English, which comes before Portuguese.

   This, that and the other (talk) 23:42, 7 December 2023 (UTC)

   @This, that and the other @Theknightwho This text works for me. Let me know whether we should wait a bit to hear from others before making the change. The only reservation I have is whether the "any layout specified" proviso would, over time, result in pretty much the same state of inconsistency that this proposal is trying to address. Also, unless I'm misunderstanding, it would no longer be a simple bottable job to order descendants (whether individual ones, or roots of subtress). However, incremental improvements can be the subject of subsequent votes - right now we have nothing in place. Chernorizets (talk) 01:28, 8 December 2023 (UTC)

   It will be trivially bottable for entries in languages without an ordering listed at the AXX page for that language. So it will be bottable for "all but n" languages. Actually, it would technically be bottable for all languages if the bot was taught the defined orderings on AXX pages where defined. This, that and the other (talk) 02:21, 8 December 2023 (UTC)

   @This, that and the other it's noteworthy btw that there is nothing on the Wiktionary:About Latin page about subfamilies or their preferred order. Is that perhaps a to-do item in view of this vote? Chernorizets (talk) 14:15, 9 December 2023 (UTC)

   @Chernorizets certainly! This, that and the other (talk) 21:51, 9 December 2023 (UTC)

   Looks good to me! MedK1 (talk) 02:33, 8 December 2023 (UTC)

   @Chernorizets @This, that and the other @MedK1 Two things:

   1. I suggest one slight further change: otherwise alphabetically by language name or family to otherwise alphabetically by language or family name - this prevents someone from interpreting it as “(alphabetically by language name) or (family)”, since we want both to be alphabetical. I trust this is uncontroversial, since it’s our obvious intention, but we know how rules lawyers can be.
   2. We should remove the first note (about reconstruction sections), since my understanding was the revised wording was to deal with the issue that affected them.

   Theknightwho (talk) 10:45, 9 December 2023 (UTC)

   @Theknightwho the notes are not part of the proposal, and I preserved the original note for transparency. The 2nd note explicitly states that concerns about reconstruction pages are subsumed under the revised text.

   I appreciate your continued efforts to improve the language of the proposal, but I won't be making further edits. I feel uncomfortable continuing to do that for an active vote with a dozen or so people who have already voted. I believe the proposal is clear enough as is, but if that proves not to be the case, I trust future votes will address that. Chernorizets (talk) 12:39, 9 December 2023 (UTC)

   I don't see why anyone could possibly be opposed to the 1st change (you have my full support). It's unfortunate that these votes are super bureaucratic though huh. MedK1 (talk) 15:02, 9 December 2023 (UTC)

   @This, that and the other I've edited the proposed change to the second sentence, since we don't want to imply borrowings can't have indentations below them. Theknightwho (talk) 23:41, 7 December 2023 (UTC)

   The proposed change sounds good to me. Whoop whoop pull up ^{Bitching Betty ⚧️ Averted crashes} 08:05, 8 December 2023 (UTC)

   I have just made the workshopped version of the wording (given by @This, that and the other in his comment above starting with "For clarity for those I pinged above") "official" in this proposal. Chernorizets (talk) 21:23, 8 December 2023 (UTC)

   @Victar language-specific considerations are now explicitly a part of the proposed text - let me know if that addresses your original concern. Chernorizets (talk) 12:42, 9 December 2023 (UTC)

2. Oppose I don't see a reason for this. Borrowed terms already use the "bor=1" argument and it adds an arrow before them. I think this proposal, if realized, would create more confusion than there is right now. If you think that the arrow is ambiguous, maybe a hint should be added somewhere instead. --GareginRA (talk) 02:22, 17 December 2023 (UTC)

To be clear, I don't oppose universal ordering in general. It's just that I'd prefer alphabetic ordering for all terms, unless you can come up with a better visual representation of direct descendants. --GareginRA (talk) 02:32, 17 December 2023 (UTC)

1. Abstain I will ignore this for descendants of descendants - I will list them under the intermediate language. There is also a problem with words that might be borrowings or might be mere cognates - and indeed some words are inherited in some senses and borrowed in others, e.g. Pali dhātu f is inherited from 'Sanskrit' (but partly a semantic loan) while dhātu m is a loan. For stability, we want to keep Pali and the mainland SE Asian languages close. --RichardW57m (talk) 10:38, 6 December 2023 (UTC)

   @RichardW57m it may not be obvious from the proposal, but my intent is to preserve the current ability to list descendant languages using either {{desc}} or {{desctree}}. So say a word has an Ottoman Turkish and a Persian descendant, and the Ottoman Turkish word in turn has a Turkish descendant. If we use {{desctree|ota}}, we'd get:

   • Ottoman Turkish
     • Turkish
   • Persian

   That's totally fine, even though P comes before T. The ordering guidance applies at the top level, where you're listing the various descendants using {{desc}} or {{desctree}}. Since Ottoman Turkish and Persian are in the recommended order, the whole tree works. That applies to any nested descendant-of-descendant subtree. Chernorizets (talk) 12:06, 6 December 2023 (UTC)

   @Chernorizets: The trouble, alas, is that your words do not preserve it. --RichardW57m (talk) 12:24, 6 December 2023 (UTC)

   @RichardW57m the proposal has been updated with some changes proposed in the discussion under the 2nd Oppose vote - I hope it fixes the issues with the old wording. Chernorizets (talk) 21:27, 8 December 2023 (UTC)

Abstain I agree with the spirit of the proposal, but per RichardW57m the wording needs to be clarified to allow sorting by not just language names but also language family names, i.e. everything at a single level within a tree should be alphabetized. -- King of ♥ ♦ ♣ ♠ 17:59, 6 December 2023 (UTC)

@King of Hearts I've updated the proposal based on suggestions from a few users (see the thread under the second Oppose vote) - let me know if that addresses your concerns. Thanks! Chernorizets (talk) 21:25, 8 December 2023 (UTC)

1. Abstain — It seems fine, but I'm not sure I understand the current treatment of descendants and inherited forms well enough to make an informed decision. Cnilep (talk) 03:50, 15 December 2023 (UTC)

Passed 17-2-2. Pinging @Theknightwho, Thadh to make the change to WT:EL. Thanks! Chernorizets (talk) 01:55, 4 January 2024 (UTC)

 Done (my username starts with "Th", so I guess it counts) This, that and the other (talk) 09:00, 6 January 2024 (UTC)

@This, that and the other Thank you! :-) Chernorizets (talk) 10:35, 6 January 2024 (UTC)

I'm kinda late to the discussion, but I support the decision very much. It was very wise to conclude it, so it's not gonna be no need to wonder later about the order of the descendant lists. The Proto-Slavic descendants seem ok, and Proto-Finnic stuff looks nice, but wait a second... Oh no, I see the Old Norse! Did I rightly understood that in the descendants of Old Norse, the Danish should be listed first? And the Old Danish must be some place in the middle (like it already does) because it starts with "O"? In this case, it is a lot of work to do. Luckily, i already used to solve some of this Danish alphabetical discrimination before anyway, but not much. I wonder why it occured this way that the Danish descendants are almost never in the alphabetical order. Anyway, this voting decision should require to be communicated to other editores who are doing the Old Norse descendants stuff, like @Leasnam, Eiliv, Mårtensås, PotatoKing147, Supevan, etc. Meanwhile, I go and hit the hay. Tollef Salemann (talk) 19:14, 1 February 2024 (UTC)

And yeah, also Faroese and Elfdalian are always after Icelandic by some reason. Tollef Salemann (talk) 19:20, 1 February 2024 (UTC)

Hey you pinged me, so I suppose I might as well give my two cents on this situation. I'm fairly new to editing Wiktionary or Wikipedia pages, but I was always fascinated by Old Norse and the decedents of it. But I did notice that it was a vast majority of the time the case that the order (from top of bottom) was Icelandic, Faroese, Norwegian (nynorsk, then bokmål), Swedish and Danish, which I found a bit strange, but I didn't think too much of it and thought it was just an invisible split between the decedents of the West-Norse and East-Norse languages and whichever was the closest to those languages (with Norwegian being the last language in this regard and Icelandic the first), and... then I guess the same goes for Swedish and Danish, but for East-Norse. So I guess another argument could be to have an alphabetical split like that, but that's gonna probably cause a lot of unnecessary struggle, so I would rather support an alphabetical split between the Old Norse descendants. PotatoKing147 (talk) 20:16, 1 February 2024 (UTC)

They are below Icelandic simply because they are further east. The order is west to east. Eiliv / ᛅᛁᛚᛁᚠᛦ (talk) 22:14, 1 February 2024 (UTC)

That's weird, why not the opposite direction? I'm not gonna order the Norse descendants if somebody is against it tho. It's a work for a week I guess. Tollef Salemann (talk) 06:12, 2 February 2024 (UTC)

It can be described as language consideration, but I don't find such a rule on the Old Norse layout description. Tollef Salemann (talk) 07:19, 2 February 2024 (UTC)