Wiktionary:Votes/2020-07/Removing letter entries except Translingual

Removing letter entries except Translingual
Voting on:


 * Removing language-specific letter entries from the mainspace. That is, any non-Translingual entry preceded by the L3-header  and/or the header  (or derivative). This vote does not affect letter entries in various sign languages.


 * Moving all these entries to appendices, depicting the alphabet of the language and any useful additional information (pronunciation, origin, inflection etc.). The layout of these appendices are left to the respective communities' discretions.


 * Updating the CFI to not include such letter entries in the mainspace.

Schedule:
 * Vote starts: 00:00, 13 October 2021 (UTC)
 * Vote ends: 23:59, 4 November 2021 (UTC)
 * Vote created: Thadh (talk) 07:58, 31 July 2020 (UTC)

Discussion:
 * [[Image:Wikt rei-artur3.svg|20px]]  (see there for previous discussions)
 * [[Image:Wikt rei-artur3.svg|20px]] 

Support

 * 1)  Thadh (talk) 05:15, 13 October 2021 (UTC)
 * 2)  &mdash; surjection &lang;??&rang; 09:52, 13 October 2021 (UTC)
 * 3)  Even if we could technically display these full entries (we can't) and even if browsers wouldn't buckle under the huge page sizes (they would), is there seriously any value in having so many entries at (e.g.) o? Who is actually helped by having "A letter in the English alphabet", "A letter in the Norwegian alphabet", "A letter in the Portuguese alphabet", etc.? Imagine how long these pages would be if we fully documented all written languages that use the Latin alphabet! It could be useful to include a navbox-style series of footers that directs users to the all the letters of these various scripts in alphabetical order because that will be different for some languages (e.g. Ñ and ñ in Spanish or İ and ı in Turkish). —Justin ( koavf ) ❤T☮C☺M☯ 03:53, 14 October 2021 (UTC)
 * Removing the non-translingual letter entries would be of minimal help with the OoM errors; as detailed in AG202's oppose vote below, the non-translingual letter entries make up only a small fraction of the total sizes of the letter pages (and this becomes an even smaller fraction if, as suggested on the talk page in response to my concerns regarding the derived-terms sections for the non-translingual letter entries, these sections are kept on the pages and moved to the translingual letter entries rather than being excised as part of the non-translingual letter entries), and many of the pages running into these errors aren't letter pages at all. My preferred option for dealing with the Lua OoM errors would be to split huge pages into several smaller pages (like we already do - for example - with the Appendix:Unicode pages for blocks with thousands of character assignments).  As for the second "problem" you list - "...and even if browsers wouldn't buckle under the huge page sizes (they would)" - that won't be the case for any reasonably-modern browser on any machine from the past decade and a half or so.  Chrome and Firefox both do just fine at handling pages far huger than our letter pages will ever get, and all the other browsers see negligible use.  Are we trying to cater to the two people in the world who still use Internet Exploder? Whoop whoop pull up Bitching Betty ⚧ Averted crashes 16:56, 21 October 2021 (UTC)
 * (Whoops, forgot to ping with that comment.) Whoop whoop pull up Bitching Betty ⚧ Averted crashes 02:00, 25 October 2021 (UTC)
 * Thanks for the ping. Re: browsers: I think you have a very Western view on this. Please try loading these pages with a KaiOS flip phone in India or Nigeria. Re: splitting pages: how would we decide when and where to split? Do you have a proposal on what that would look like? E.g. O/1, O/2, etc.? —Justin ( koavf ) ❤T☮C☺M☯ 06:25, 25 October 2021 (UTC)
 * Re: re: splitting pages, I would suggest something like A/Translingual...German, A/Herero...Maltese, A/Monegásque...Turkish, A/Vietnamese...Zulu, or the like (language labels in these links not necessarily ones that actually currently exist on that particular page; this is just a mockup), or maybe A/Languages: A-F, A/Languages: G-M, A/Languages: N-S, A/Languages: T-Z (in the latter case, the translingual and English entries would simply stay at the main letter page, with all the other language-specific entries - regardless of type of speech - being moved to the bogosubpages). Re: re: browser limitations, is there anything to indicate that a significant number of Wiktionary users are actually accessing the site by such low-capability means?  And do we really want to hold ourselves back that severely to cater to such a low common denominator?  (And and - and speaking here as someone who used a KaiOS flip phone until a few days ago - would it even be possible to do so?  MediaWiki no longer supports rendering even on older desktop web browsers - and even an old, obsolete, barely-functional desktop browser [like IE] is still a giant leap ahead in capabilities and functionality compared to a flip-phone browser.) Whoop whoop pull up Bitching Betty ⚧ Averted crashes 14:06, 25 October 2021 (UTC)
 * Woah, wait a second. I still think that there needs to be a serious revamp of the pages, and catering to those with underserved Internet services and not-as-powerful systems would be beneficial for the website as a whole. I just didn't think that removing letters was the way to go about it (short-term half-solution for a much longer-term problem). AG202 (talk) 14:30, 25 October 2021 (UTC)
 * Do we want to cater to lo-powered browsers in the developing world/Global South/Third World/etc.? Yes, of course. I'm not certain of what metric we'd want to use (i.e. I mean, I don't know what percentage of Internet users we would want to design around excluding in favor of newer features; I'm inclined to say 0.1% or even 0.01%) but I'm not comfortable deliberately making policies that exclude 1%+ of Internet users: if they are excluded then by definition, they won't be Wiktionary users in the first place! —Justin ( koavf ) ❤T☮C☺M☯ 03:28, 27 October 2021 (UTC)
 * ...To change the subject slightly, waddya think of my method-of-splitting skeleton proposal? Whoop whoop pull up Bitching Betty ⚧ Averted crashes 00:54, 29 October 2021 (UTC)
 * Could work, seems sensible. No solution will be perfect, but this is a good idea. —Justin ( koavf ) ❤T☮C☺M☯ 02:09, 31 October 2021 (UTC)
 * (Just to clarify, are you still voting in support of nuking the non-translingual letter entries from mainspace, or are you more in favor of keeping them in mainspace and splitting the big pages up into manageable chunks [as per my proposal]?) Whoop whoop pull up Bitching Betty ⚧ Averted crashes 03:40, 31 October 2021 (UTC)
 * I am still in favor of my original proposal but I think yours is acceptable. —Justin ( koavf ) ❤T☮C☺M☯ 03:46, 31 October 2021 (UTC)
 * @Koavf Thanks for the explanation about browsers! And honestly, (since Nigeria was mentioned and I am Nigerian, I feel obligated to respond) this is a problem that's already evident and is not solved at all by moving letters. As mentioned in my comment under "Oppose", the letter pages have Lua errors, not because of the letter entries, but because of other entries. mi has a Lua Memory Error with no letters after all, and then letter entries tend to be shorter regardless, so the pages buckle under other content. So overall, if we're really trying to cater to those in areas with underdeveloped Internet, the website as a whole will need a serious revamp. AG202 (talk) 08:42, 25 October 2021 (UTC)
 * Your point about some non-letter pages being too long is well-taken. —Justin ( koavf ) ❤T☮C☺M☯ 08:51, 25 October 2021 (UTC)
 * I'm open to the footers idea, could you maybe give more detail about that if at all possible? AG202 (talk) 05:01, 14 October 2021 (UTC)
 * Just a simple navbox that is something like "English alphabet" and has A/a, B/b... all in one or two rows, followed by "Afar alphabet", "Afrikaans alphabet", etc. Still a lot of templates but far fewer and taking up much less space. —Justin ( koavf ) ❤T☮C☺M☯ 17:29, 16 October 2021 (UTC)
 * : Category:Latin script languages = 3,637. —Suzukaze-c (talk) 05:05, 14 October 2021 (UTC)
 * 1)  --Numberguy6 (talk) 01:12, 15 October 2021 (UTC)
 * 2)  in principle, but with the provisos that (a) I see no reason not to include pronunciation, derived terms etc. in the main Translingual entry, grouped by language section in a bulleted list (see User:This,_that_and_the_other/translingual), and (b) an exception should be made for letters that only occur in one language (e.g. the letters of the Armenian alphabet, or ß). This, that and the other (talk) 08:50, 15 October 2021 (UTC)
 * I don't understand why Armenian letters and the Eszett are more worthy of a non-Translingual entry than other letters. Also, if you want to add all pronunciations in all languages to the Translingual section, its pronunciation section will become enormous. Thadh (talk) 09:34, 15 October 2021 (UTC)
 * I am suggesting that single-language letters could receive an entry under their language L2 header instead of Translingual, because the letters are not "translingual" in any sense of the word. This is how, for instance, Խ is currently formatted. It lacks a Translingual section and has the letter entry under Armenian.
 * And yes, the pronunciation section will become large, but a collapsible box (like we do for translations) can be used, as it already is at . I don't see a good reason why that information needs to be pushed into an appendix. However, I'm not terribly fussed about this aspect of the proposal. This, that and the other (talk) 11:11, 15 October 2021 (UTC)
 * I mean, I guess it doesn't really matter whether language-specific letters are translingual or language-based, but I would think an English text could say something like "The Armenian letter Խ is used to...", which would render the letter Translingual? I'm not sure, I don't know what our customs on this are; I guess this is something to be discussed either after the vote or on the talk page, since it's part of execution. Thadh (talk) 11:19, 15 October 2021 (UTC)
 * @Thadh@This, that and the other Armenian should be Translingual so long as Old Armenian is a separate language header, IMO. Tibidibi (talk) 03:24, 16 October 2021 (UTC)
 * Yes, I'm not opposed to that. All I'm saying is that we should retain the flexibility to not use Translingual in cases where a letter belongs to only one Wiktionary-recognised language, just as is proposed for sign languages. This, that and the other (talk) 03:26, 16 October 2021 (UTC)
 * 1) . I'd also support a similar solution for famous surnames . Imetsia (talk) 18:55, 16 October 2021 (UTC)
 * 2) . —Mahāgaja · talk 13:47, 18 October 2021 (UTC)
 * 3) Heavy-hearted  for scripts that are used in a lot of languages. I find that the benefits outweigh the negatives. — Ungoliant (falai) 10:38, 28 October 2021 (UTC)
 * Is this a provisional support vote? AG202 (talk) 13:01, 28 October 2021 (UTC)
 * on entries for letters that are used in fewer than 25 languages. — Ungoliant (falai) 22:54, 28 October 2021 (UTC)
 * Got it, thanks! AG202 (talk) 22:56, 28 October 2021 (UTC)
 * I object to partial votes like this, which allow an individual's desires to win the day when they happen to be the deciding vote. I think the closer of this vote would be justified in ignoring such stipulations. —Μετάknowledge discuss/deeds 01:33, 29 October 2021 (UTC)
 * 1) . Most of this content is redundant and not very useful. Perhaps there can be exceptional cases but I think removing them would do more good than bad. Vininn126 (talk) 13:55, 28 October 2021 (UTC)
 * 2) . Letters aren't words, and our current approach is unsustainable. —Μετάknowledge discuss/deeds 20:46, 28 October 2021 (UTC)
 * Letters are words - single-letter words. Whoop whoop pull up Bitching Betty ⚧ Averted crashes 00:54, 29 October 2021 (UTC)
 * I agree. Letters are words. If this sounds like a poetic justice,
 * at least morphemes in agglutinative languages in three digits are. Stand-alone those morphemes  make as much sense as any letter does.
 * "muvaffakiyetsizleştiricileştiriveremeyebileceklerimizdenmişsinizcesine"
 * As though you are from those whom we may not be able to easily make into a maker of unsuccessful ones.
 * As far as I know each word in sentence above has a meaning.
 * Yes, I'm aware you're voting for the letters and not for the morphemes.
 * Flāvidus (talk) 01:00, 3 November 2021 (UTC)
 * reluctantly. I hope each letter entry will have a link to an appendix listing the languages that use a letter so that the appendices are findable.... And I hope that appendices will be able to include etymological information, pronunciation, and usage notes for each letter. Andrew Sheedy (talk) 01:05, 29 October 2021 (UTC)
 * Changing vote to an abstention. Andrew Sheedy (talk) 20:18, 1 November 2021 (UTC)
 * 1)  Allahverdi Verdizade (talk) 22:30, 29 October 2021 (UTC)
 * 2) . The information in these sections is not worth preserving because A) there is usually very little of it and B) it's scattered across dozens of pages. The most obvious alternative to me is a language-specific appendix (e.g. Appendix:Spanish alphabet) with the letters in order, their pronunciation, letter names, and any notes. There's nothing to stop someone from creating/improving these right now. Ultimateria (talk) 17:13, 30 October 2021 (UTC)
 * @Ultimateria @Thadh Since this vote is likely to pass, I'm still wondering what's going to happen to obsolete, dated, & more specific letters (see below for a few), since they wouldn't fall under the typical alphabetical order, and would have to be addressed more specifically. I also came across this entry,, and am a bit concerned as to how the declension entries would loop back to an appendix page + how the declensions would appear for each letter. AG202 (talk) 18:20, 30 October 2021 (UTC)
 * The Pali entry is a noun, not a letter. As for obsolete forms, I imagine they will be given in the same appendices as the other letters, just under a separate header. Thadh (talk) 02:06, 31 October 2021 (UTC)
 * The Pali entry really looks like a letter, maybe treated as a noun just as a convention. ·~   dictátor · mundꟾ  14:42, 31 October 2021 (UTC)
 * What attribute of a noun does it lack? Also, it doesn't specifically mean 'the Devanagari letter र'.  Note that the second quotation establishes that the word  gets inflected.
 * Letters may be used as nouns ("One a and two bees"), but these should be verifiable like any noun. See the first bullet point in the vote's premise: "That is, any non-Translingual entry preceded by the L3-header ===Letter=== and/or the header LANG (or derivative)." Thadh (talk) 16:00, 31 October 2021 (UTC)
 * @Thadh it should be under the letter header, and I think that's something done by the person who made the entry (I recall this exact example being mentioned before). And then if we can do letters as nouns like that, then what's stopping people from making an entry for the "noun" a for all Latin-script languages? Isn't this precisely what sparked the vote in the first place? Would they not be nouns in this example: "Which letters are in the word 'beet'?" — "One b, two e's, & one t." That construction is more common than the letter names, I'd say, and would be verifiable under CFI, so at that rate, I'm more confused as to what this is preventing. AG202 (talk) 17:00, 31 October 2021 (UTC)
 * I feel like we're discussing the same things over and over again. This question in particular has been answered on the talk page, but here's a summary: A letter is verified by any instance of using it in a particular language. The word "hello" alone could be a quote for four different English letters. Nouns, on the other hand are much trickier, because they have to be used in a running text as nouns. "b" isn't an English noun, because the name of the letter is "bee" in English. Similarily, in Greek you have άλφα, in Yiddish you have בית, and in Old East Slavic you have землꙗ. These are all nouns, not letters, and they function differently in a language. Thadh (talk) 17:48, 31 October 2021 (UTC)
 * Moving this to the discord server. AG202 (talk) 18:00, 31 October 2021 (UTC)
 * 1)  Though I think we will be disappointed.  As I write the name of the letter as 'b', with the apostrophes, what page should that noun be on?  If I wrote, 'I was staring at a bee', I'm talking about looking at an insect, not a letter. RichardW57 (talk) 21:16, 1 November 2021 (UTC)
 * 2) . --Fytcha (talk) 10:21, 4 November 2021 (UTC)
 * 3)  Pious Eterino (talk) 10:30, 4 November 2021 (UTC)
 * 4)  Kutchkutch (talk) 11:54, 4 November 2021 (UTC)
 * 5)  While the ideal solution would be to have letter entries present in one long page, the issue of lua errors causing long entries to break makes that infeasible. The proposition fixes that major issue, and any additional issues or ambiguities that are caused can be addressed as they come up.  – Guitarmankev1 (talk) 17:34, 24 November 2021 (UTC)

Oppose

 * 1)  for now. While I appreciate the work that Thadh has done with answering the questions that I’ve had, there are still so many issues that need to be address before I can comfortably vote for such a sweeping measure across all languages. To the question of whether or not letters are dictionary content: why not just look at common dictionaries. For the languages that I’m most familiar with:
 * 2) * Oxford Dictionary has an entry for e
 * 3) * Merriam-Webster has an entry for h
 * 4) * Collins has an entry for j
 * 5) * Dictionary.com has an entry for q
 * 6) * La Real Academia Española has an entry for ñ
 * 7) * L’Académie Française has an entry for o
 * 8) * 표준국어대사전 (The Standard Korean Language Dictionary) has an entry for ㅊ
 * 9) * Most Yorùbá dictionaries have entries for their letters or an explanation
 * 10) * Dicio (Portuguese) has an entry for b
 * 11) * El Diccionari de la llengua catalana de l'Institut d'Estudis Catalans has an entry for u
 * So to me, it definitely seems like they are dictionary content to major dictionaries, so why would we cause ourselves to have less information on entry pages? (Let alone the fact that they are lemmatized) There’s so much information that goes along with letters as well, as they are words in their own rights, such as pronunciations, etymologies, ᆢ|historical information, mutation, inflection, derived terms, gender (which should be included for gendered languages already if applicable), form information, references, usage notes, obsolete, dated, or more specific letters found outside an alphabet, and more that we would have to put on editors to make a (clean) appendix for, meanwhile it works in the way that it is right now. Also, with languages like American Sign Language where their letters are under the Letter header and there’s important production information, how would that fit cleanly into an Appendix page? (Vote updated to exclude Sign Languages) With more ancient or extinct or even newer languages/scripts as well, it is very helpful to have the letter as its own entry/page. I'm sure there's much much more that I've missed while doing my own research as well that hasn't even been touched. Therefore, while I wish that there were more discussion, I just can’t bring myself to vote to force editors to move all their work and create new pages & systems unnecessarily, let alone all the broken templates and links. And before anyone mentions Lua Memory Errors, this is not the solution for that, if you look at many of the entries like o or e, most of the entries are in fact not letters, but other POSs, and then, mi, for example, that has no letters has the same issue. Thus, deleting letters & moving them to translingual is not a solution, but a wide-sweeping move that wouldn’t fix that issue at hand to begin with, a short-term solution to a long-term problem. Also, I feel like the solution of having Translingual have a link to "languages that have x letter" -> appendix of language letter is very clunky and not super clear for readers (the Appendix is a bit out of reach to begin with as well). And while this isn't as relevant, I've found it really helpful to see how letters like c change between languages and develop without having to click through tons of appendices, which is part of what Wiktionary is for. Thus, overall, I don’t feel comfortable voting for this proposal in the way that it’s formatted right now, without at least a significant discussion from other language communities, at least if we're really striving for "All words in all languages". AG202 (talk) 00:02, 13 October 2021 (UTC)
 * For what it's worth, I did the work of removing every instance of a ==Letter== header from a and the Lua memory errors still start at the exact same language, see: User:AG202/Sandbox2, so after this vote, there needs to be actual work made towards fixing this issue. Pinging @Koavf, @Andrew Sheedy, & @Whoop whoop pull up for clarity/information's sake. AG202 (talk) 15:23, 4 November 2021 (UTC)
 * There's still lots of letter name entries on your sandbox page. Are those not removed as well under this proposal? See e.g. Icelandic. Fytcha (talk) 15:43, 4 November 2021 (UTC)
 * @Fytcha Those are explicitly not being removed by this current proposal (were previously included but then removed if I recall correctly). AG202 (talk) 15:47, 4 November 2021 (UTC)
 * Late comment, for the record. Catalan IEC dictionary "has an entry for u". True, it is the noun for letter u or U. This happens with vowels. There is no entry for consonants, but its names:, etc. Vriullop (talk) 08:16, 10 November 2021 (UTC)
 * 1)  Urszag (talk) 03:19, 14 October 2021 (UTC)
 * 2)  per AG202, and also because the language-specific letter entries are the source of a great many language-specific derived terms, which would be left hanging (with nowhere to list them as being derived from) were the language-specific letter entries to be deleted. Whoop whoop pull up Bitching Betty ⚧ Averted crashes 03:13, 15 October 2021 (UTC)
 * 3)  Svartava2 (talk) 13:12, 17 October 2021 (UTC)
 * 4)  Mölli-Möllerö (talk) 16:46, 26 October 2021 (UTC)
 * 5) . It’s sad Thadh wants to make this project less informative and user-friendly.  ·~   dictátor · mundꟾ  19:23, 27 October 2021 (UTC)
 * Moving information from one place to another does not make the project less informative. Less user-friendly perhaps, but I question this conclusion given the bloat in single-letter entries. — Ungoliant (falai) 10:41, 28 October 2021 (UTC)
 * Will the Translingual entries contain links to the Appendices? The system will not become less user-friendly if a Translingual entry contains links to all the languages’ letters, but I think that is not going to happen. ·~   dictátor · mundꟾ  16:55, 28 October 2021 (UTC)
 * That was the idea... (either in the Translingual entry or at the top of the page). Please read the proposal and discussions. Thadh (talk) 17:01, 28 October 2021 (UTC)
 * I am afraid the vote description is not elaborate. Anyway, what is the idea now? ·~   dictátor · mundꟾ  18:20, 28 October 2021 (UTC)
 * That is still the idea. The implimentation is not wholly set in stone, but see my mockups at User:Thadh/Translingual/a, the in that entry linked to User:Thadh/Appendix:List of languages using the letter "a", User:Thadh/Appendix:List of Afar letters and User:Thadh/Appendix:List of Welsh letters. Thadh (talk) 18:25, 28 October 2021 (UTC)
 * 1)  AntisocialRyan (talk) 16:41, 2 November 2021 (UTC)
 * 2) .--Tibidibi (talk) 16:06, 3 November 2021 (UTC)
 * Maybe apply it only to scripts used by more than ten languages. I agree that having several thousand entries for the Latin alphabet letter "q" would be dumb. Tibidibi (talk) 16:07, 3 November 2021 (UTC)
 * 1)  -Solarkoid (talk) 07:48, 4 November 2021 (UTC)
 * 2)  Oníhùmọ̀ 23:12, 4 November 2021 (UTC)
 * 3)  Primarily regarding the effect this will pose on the Yorùbá entries and the role letters play in other Yorùbá dictionaries.- Oniwe (talk) 23:30, 4 November 2021 (UTC)
 * 4)  [late] blanket removals, e.g. entries like Thai  or Khmer . It makes sense to remove Latin, Cyrillic entries. However, some letter entries perhaps should be considered on their merits, e.g. Russian letters,  and  are not so straighforward in their readings or usage. Also digraphs, like Czech . --Anatoli T. (обсудить/вклад) 22:13, 21 November 2021 (UTC)
 * Well, we're now free to add the Northern Thai, Northern Khmer, Pattani Malay and NE Thai letters ด.   And we currently have seven non-translingual letters ch. --RichardW57 (talk) 00:24, 22 November 2021 (UTC)
 * I would think differently if the information about (e.g. Thai) letter names (both alternatives) and how those names are spelled, consonant classes were throroughly copied to appendices but it's not the case, so. I don't want valuable info to be lost despite many entry duplications. [Wiktionary:Thai romanization] is almost there but I need letter names (perhaps in other appendices). --Anatoli T. (обсудить/вклад) 01:01, 22 November 2021 (UTC)
 * The appendices were to be created after the vote passed. --16:02, 22 November 2021 (UTC)

Abstain

 * 1)  while I think about this. I agree with not having three thousand L2 sections for all of the languages that use a; preserving the letter names and their pronunciations, etc (e.g., notice how much content ß has) in appendices seems like a better approach. I agree with "This, that" that if a letter is truly only used in one language, it should just have that language's L2 and not "Translingual", but this is rarer than suggested: for example, Armenian letters are also used by Old Armenian, and ß is also used in other German languages like Bavarian and Zipser. - -sche (discuss) 21:09, 16 October 2021 (UTC)
 * Also non-German languages of Germany. It was formerly used in Sorbian, for example. —Mahāgaja · talk 13:50, 18 October 2021 (UTC)
 * , seeing as the proposal seems to be popular. I wish Thadh good luck with the vote, even tho’ I myself am not inclined to support it. I know not how such a big-scale change could be implemented… ·~   dictátor · mundꟾ  11:47, 30 October 2021 (UTC)
 * 1) . I've decided to withdraw my support, following the comments made on the Talk page about this not actually solving the Lua memory issue, but I'm refraining from opposing, because I can still see some benefits with this path. I can imagine that having 3,000 entries on a single page in the future would be a bit ridiculous. Andrew Sheedy (talk) 20:18, 1 November 2021 (UTC)
 * 2) . DonnanZ (talk) 23:34, 4 November 2021 (UTC)

Decision

 * No consensus: 16-10-4 [per, ignoring 's stipulational vote] —Svārtava [t•c•u•r] 03:32, 5 November 2021 (UTC)
 * You misinterpreted my statement. I said that his stipulations should be ignored, not his vote. (But at 17–10–4, it would still be no consensus.) —Μετάknowledge discuss/deeds 05:41, 5 November 2021 (UTC)