Wiktionary:Beer parlour/2017/June

From Wiktionary, the free dictionary
Jump to navigation Jump to search
discussion rooms: Tea roomEtym. scr.Info deskBeer parlourGrease pit ← May 2017 · June 2017 · July 2017 → · (current)

Last remaining private use area characters[edit]

The following pages have private use characters: , proposition, 슴새, bên, 배다, xǔxǔ, and 癩 (a redirect). Can we clean these up? DTLHS (talk) 05:06, 1 June 2017 (UTC)Reply[reply]

Out of curiosity, did you check all namespaces, or just the mainspace? Just the other day, I changed a PUA character in a sortkey of Module:languages/data3/t! - -sche (discuss) 05:10, 1 June 2017 (UTC)Reply[reply]
Just mainspace. DTLHS (talk) 05:15, 1 June 2017 (UTC)Reply[reply]
@-sche Here's the entire site if you want to see it. DTLHS (talk) 05:36, 1 June 2017 (UTC)Reply[reply]
Thanks! The uses in userspace and some other places can probably be left as-is, but e.g. the uses in Template:IPAsym looks like ones we should check. In one case, I see that the character has since been added to Unicode. - -sche (discuss) 18:26, 1 June 2017 (UTC)Reply[reply]
All done. Wyang (talk) 07:33, 1 June 2017 (UTC)Reply[reply]

Conventions for Egyptian[edit]

@Furius, Hyarmendacil, Strabismus, CAmbrose I’ve been doing some work with Egyptian and have come across a number of problems that aren’t yet addressed or could do with changing in WT:About Egyptian, so I thought I’d ask about them to try to come to some sort of consensus before going ahead with implementing anything unilaterally. (Unfortunately I think all the Egyptologically inclined editors I pinged are inactive, but I figured it’s worth a try.) The proposals I would make are these:

  • Distinguish between s and z instead of merging them both into s like we (nominally) do now. The phonemes merged by Middle Egyptian, but they were still separate in Old Egyptian, and most authors still make the distinction, from Allen to the Wörterbuch. Even Faulkner’s dictionary of Middle Egyptian puts the variants with z in parentheses where appropriate. On Wiktionary, Egyptian covers Old Egyptian as well as Middle, so that separating z out makes more sense, and there are already a few entries that have z in defiance of the current policy.
  • Change 3 to Ꜣ in transliteration. This is the correct Unicode codepoint dedicated to Egyptological aleph, and font support has become good enough that I think most people can see it. (Is this actually the case?) The current 3 wreaks havoc with the linkify function in templates, which converts it to an unlinked superscript 3 whenever it comes at the end of a word.
  • Change ˤ to ꜥ in transliteration. Not so important, just a switch to the correct codepoint now that it’s probably supported for most people. (Again, can anyone confirm/deny?)
  • Standardize the hieroglyph Z4 (the two dual strokes
    ) and the nisba adjective ending to be a variant of j (
    ) rather than y (
    ). (Right now it is not defined either way.) Both conventions are common — Faulkner and Hoch call it y, while Allen, Loprieno, and the Thesaurus Linguae Aegyptiae call it j — but the convention in most of our entries was to call it j, presumably because most of our editors were reading Allen. I’m proposing that we standardize what already seemed most common.
  • Only use dots (.) to separate morphemes in the case of suffix pronouns etc., not with inflectional suffixes for singular, plural, etc. This is again the convention that already seems most common, presumably because it’s the one Allen follows. I’d be fine with the other common system, too, where equals signs (=) separate out the suffix pronouns and dots separate out all the other morphemes, but consensus seems to favor the one Allen uses.
  • Standardize the use of periods as dots, rather than using interpuncts (·). Again for this, I have no opinions either way, but this seems to be current consensus.

Thoughts? —Vorziblix (talk) 07:56, 1 June 2017 (UTC)Reply[reply]

I don't remember when and why our current standards were established, but it is true that nobody working on them is active any longer. I think all your points seem very reasonable, although alternate transliterations should remain as soft redirects to the standardised lemma form using {{egy-alternative transliteration of}}. —Μετάknowledgediscuss/deeds 15:16, 1 June 2017 (UTC)Reply[reply]
Most of these proposals seem reasonable; I do not know enough to comment on j vs y. I strongly support avoiding the equals sign =, which causes difficulties if provided in a template (link, etc), and possibly also if we were to devise a module that parses the content of our entries to generate something (as is done in some Thai entries, and recently proposed for Arabic), and possibly also for some re-users of our content, because of what it usually means in template syntax. - -sche (discuss) 18:37, 1 June 2017 (UTC)Reply[reply]
@Vorziblix: I can confirm that the and characters (as well as ˤ) all display on the machine I am currently using, which is a loaner that does not have obscure font support. —Justin (koavf)TCM 19:44, 1 June 2017 (UTC)Reply[reply]
I don't edit Egyptian, but and just show up as boxes for me... Andrew Sheedy (talk) 02:22, 2 June 2017 (UTC)Reply[reply]

Thanks for all the feedback. Let’s consider the use of the equals sign definitively ruled out, and the soft redirect policy makes sense. Regarding lack of universal support for and , I am inclined to think the template issues and Unicode compliance still make the changeover worthwhile, but will defer to consensus if the general opinion is otherwise. In any case I’ll wait a few more days to give more time for comments before making any changes/additions to WT:About Egyptian. —Vorziblix (talk) 04:20, 2 June 2017 (UTC)Reply[reply]

The symbols also display correctly for me. If we switch away from , remember to update that entry, which currently mentions the use of ˤ. I wonder if it would help to switch egy from using Latn as its script, to using Latinx, which I think calls on more comprehensive fonts. - -sche (discuss) 05:20, 2 June 2017 (UTC)Reply[reply]
That’s probably a good idea; the characters are in the LATIN-EXTENDED-D block, which is explicitly looked for by Latinx but not by Latn, if I’m interpreting that page correctly. —Vorziblix (talk) 20:26, 2 June 2017 (UTC)Reply[reply]
@-sche Since I can’t edit Module:languages/data3/e without admin privileges, would you be willing to make the switch to Latinx? Thanks, —Vorziblix (talk) 20:24, 21 June 2017 (UTC)Reply[reply]
 Done. - -sche (discuss) 21:24, 21 June 2017 (UTC)Reply[reply]
I am very inactive, sadly, but all of these are reasonable. The first point (s/z) is the one that is most likely to cause issues, I think. What would the new rule mean for e.g. nts? So far as I'm aware (and I confess that I have never studied old Egyptian) the alternative reading only comes into existence once z and s had merged. Would we create a new article for ntz? Furius (talk) 15:57, 2 June 2017 (UTC)Reply[reply]
The lemma form would definitely be at nts, since it derives from the fem. sing. 3p. suffix pronoun .s, which was s all through Old Egyptian; if ntz exists at all, it should just be an ‘alternative form of…’ entry, but it’s probably unneeded. In general words that first appear after OE and never had the s/z distinction would be lemmatized with ‘s’; words that had ‘z’ in OE would be at the ‘z’ variant, but with an ‘alternative form’ entry at the ‘s’ variant; and words that had ‘s’ in OE would straightforwardly be at the ‘s’ variant. That way anyone searching by ME form could always search with ‘s’, and anachronistic readings like ntz wouldn’t be necessary.
By the way, thanks for your work with verb conjugations; it helps a lot with making templates! —Vorziblix (talk) 20:26, 2 June 2017 (UTC)Reply[reply]
As further potentially useful information, at the bottom of this page there’s a discussion on the principles used by Allen for transliteration; regarding s vs. z they line up pretty well with my suggestion above: ‘If the hieroglyphic writing of a particular instance of a word has z but the original spelling had s, we transliterate the consonant as s,… but vice versa, if a particular instance has s but the original spelling had z, we transliterate the consonant as s.… We transliterate as z only if both the original spelling and the particular instance have z.’
Some of the other discussions there might also be good to consider, particularly the question of whether to hyphenate compound words, where we don’t seem to have any consistent policy. —Vorziblix (talk) 01:41, 3 June 2017 (UTC)Reply[reply]

Barring further objections, I’ll start moving ahead with standardizing entries. —Vorziblix (talk) 05:49, 5 June 2017 (UTC)Reply[reply]

Enable sitelinks on Wikidata for Wiktionary pages (outside main namespace)[edit]

Hello all,

Here's an important information about the evolution of Wiktionary sitelinks in the next weeks.

Short version: From June 20th, we are going to store the interwiki links of all the namespaces (except main, user and talk) in Wikidata. This will not break your Wiktionary, but if you want to use all the features, you will have to remove your sitelinks from wikitext and connect your pages to Wikidata.

Long version available and translatable here.

If you have any question or concern, feel free to ping me.

Thanks, Lea Lacroix (WMDE) (talk) 08:23, 1 June 2017 (UTC)Reply[reply]

@Lea Lacroix (WMDE) What do you mean by "all the namespaces"? There are many custom namespaces (Appendix, Concordance, Index, Rhymes...) Note the discussion at Wiktionary:Beer parlour/2017/May#Use "Cognate" to link between citation pages. --Vriullop (talk) 08:59, 1 June 2017 (UTC)Reply[reply]
The namespaces you mention can be stored in Wikidata as well. The namespace Rhymes, for example, has an equivalent on German Wiktionary, so it will be possible to make links between the pages. About Citations: we're going to investigate to see if it is more relevant to make automatic links with Cognate, or centralized links in Wikidata. Lea Lacroix (WMDE) (talk) 09:08, 1 June 2017 (UTC)Reply[reply]
For what it's worth, the Portuguese Wiktionary has a Rhymes namespace too, but it's virtually unused. It has only one page, which is obviously a stub: pt:Rimas:Inglês. But I suppose it wouldn't hurt to link it to our Rhymes:English through interwikis. --Daniel Carrero (talk) 13:36, 1 June 2017 (UTC)Reply[reply]
Citations pages match their connected main-namespace pages. An exception is when citations are centralized on one page; one wiki might choose to centralize citations of both "have got someone's back" and "have someone's back" on Citations:have someone's back, where another wiki might centralize them on its equivalent of Citations:have got someone's back ... but one wiki might choose to lemmatize one of those phrases in the main namespace, too, where another wiki might lemmatize the other phrase ... and hard or soft redirects might or might not exist ... much like if a certain discussion happened to take place only on Talk:have got someone's back and not Talk:have someone's back. So IMO it makes as much sense to handle Citations pages via the Colgate extension, as it does to handle main-namespace and talk pages. Interwiki links between citations pages seem not very useful, anyway. - -sche (discuss) 05:48, 2 June 2017 (UTC)Reply[reply]
I believe Cognate is the better option for Citations pages. Please see my reply at Wiktionary:Beer parlour/2017/May#Use "Cognate" to link between citation pages. --Daniel Carrero (talk) 11:11, 2 June 2017 (UTC)Reply[reply]
What about categories? Most Wiktionaries have the same category for each language, so adding each of them individually to Wikidata is a huge waste of effort. There's no difference between Category:English nouns and Category:Dutch nouns other than the language, they should be treated as the same thing. —CodeCat 18:39, 1 June 2017 (UTC)Reply[reply]
This phase is the same done for Wikipedia and other projects, so any interwiki link existing on any page will be exported to Wikidata, except user pages, talk pages and main space provided by Cognate. That means Wiktionary: pages, categories, templates, modules, etc., including custom namespaces I asked for. You have not to add them individually, it is a mass export. So Category:English nouns will continue linking to nl:Categorie:Zelfstandig naamwoord in het Engels, etc., but it will be easier to maintain it in a centralized place. Any interwiki link in one language project will appear in all wikts. Any page renamed or deleted will update interwikis immediately (as currently does Cognate). See d:Q7923975. I suppose that besides Wikipedia, Wikibooks, etc, it should appear Wiktionary with Category:English language, nl:Categorie:Woorden in het Engels and all wikt interwikis. Apart from interwiki links for Wiktionary you can access to equivalent links in other projects, if any. --Vriullop (talk) 06:53, 2 June 2017 (UTC)Reply[reply]
About CodeCat's "There's no difference between Category:English nouns and Category:Dutch nouns other than the language ..." This only applies to a small minority of categories in the English Wiktionary. Yes, if we had a centralized database of languages, we could then generate a list of "Category:(language) nouns" in the English Wiktionary, but do many Wiktionaries use a predictable system for their nouns categories, and all other categories? Their category systems might unpredictably change at some point, and some existing and future Wiktionaries might not have figured standards for categories yet. Maybe Wikidata is not the most perfect conceivable system for listing our category interwikis using little storage space but at least it seems to be as flexible as needed. --Daniel Carrero (talk) 11:20, 2 June 2017 (UTC)Reply[reply]
@Daniel Carrero: Individual Wiktionaries moving things is an argument for a central repository at Wikidata. If they get moved on the (e.g.) Dutch Wiktionary, then the links will be instantly and automatically updated at the (e.g.) Swahili Wiktionary. —Justin (koavf)TCM 15:53, 2 June 2017 (UTC)Reply[reply]
It just occurred to me that we would probably want Category:English language directly connected to d:Q1860, and similar for the main category of other languages. Would this be done as part of this change? —CodeCat 19:55, 11 June 2017 (UTC)Reply[reply]
d:Q7923975 is the concept "Category:English language" in multiple Wikimedia projects, including Wikiversity, Wikibooks and Wikisource. Its value "category's main topic" is d:Q1860. --Daniel Carrero (talk) 20:01, 11 June 2017 (UTC)Reply[reply]
Hmm, I can't say I understand why there's this distinction. Wikipedia's page w:English language is really not about anything different than our Category:English language. —CodeCat 20:02, 11 June 2017 (UTC)Reply[reply]
Probably just to store interwikis for the same category in multiple Wikipedia languages: w:en:Category:English language, w:nl:Categorie:Engels... You get the point. This, in addition to the "Category:English language" in Wikibooks, Wikisource etc. as mentioned above. --Daniel Carrero (talk) 20:06, 11 June 2017 (UTC)Reply[reply]
But it has a much more prominent role on Wiktionary than on any of those other sites. It's our main page about English, just like w:English language is Wikipedia's. —CodeCat 20:11, 11 June 2017 (UTC)Reply[reply]
See for example English Wikibooks. b:Subject:English language is connected to d:Q1860 and b:Category:Subject:English language to d:Q7923975. Accordingly, Wiktionary:About English should connect to Q1860 and Category:English language to Q7923975. --Vriullop (talk) 12:01, 12 June 2017 (UTC)Reply[reply]
I agree with Vriullop. Our Category:English language is pretty awesome, with the description full of diverse English-related links, but its main purpose is still just being the main category about English. Wikidata controls interwikis, and the category interwikis should probably be between categories only. --Daniel Carrero (talk) 12:32, 12 June 2017 (UTC)Reply[reply]
Is there a discussion page or something else about these "category" items, that might explain their purpose and thus help decide whether Category:English language should be in there? —CodeCat 20:13, 11 June 2017 (UTC)Reply[reply]
@CodeCat: I don't know about any discussions, but as I'm sure you know, apparently anything that exists in separate Wikipedia pages also may have separate Wikidata items for the sake of storing interwikis.
--Daniel Carrero (talk) 14:07, 14 June 2017 (UTC)Reply[reply]
To be more exact, apparently any page in some Wikimedia projects can have its own Wikidata item to store interwikis. For example, d:Q30237873 is the recent Wikinews article "Theresa May's Conservative Party wins UK election but loses majority, leaving Brexit plan in question". --Daniel Carrero (talk) 06:58, 19 June 2017 (UTC)Reply[reply]

Cognate & redirects[edit]

Hello again,

Several persons mentioned the idea of Cognate linking to redirection pages. This is a complex issue that should be decided with a consensus of different languages communities. To complete the discussions that have been running on Phabricator and have more points of view, I created a discussion topic here. Feel free to add a comment. Thanks, Lea Lacroix (WMDE) (talk) 13:13, 1 June 2017 (UTC)Reply[reply]

lexicographic approach to learning[edit]

One the main uses of a lexicographic resource is that of educational learning, therefore it would enrich Wiktionary to create a discussion room to ask for advice from advanced users regarding the learning process itself. The feedback would enable discovering current weaknesses and so improve Wiktioanry on the whole. --Backinstadiums (talk) 15:07, 1 June 2017 (UTC)Reply[reply]

But it still leaves the learning process with an excessively narrow focus on people like us. We are already pretty good at taking ourselves as model users. We need more than a convenience sample of active, interested users of Wiktionary. Any thoughts on how to get that? DCDuring (talk) 16:45, 1 June 2017 (UTC)Reply[reply]
@Backinstadiums, can you give an example of what kind of discussion you would expect to see in such a page? — Ungoliant (falai) 01:31, 2 June 2017 (UTC)Reply[reply]

Personal advice from those who are already fluent, what they'd do differently knowing what they know now, mistakes to be avoided, grammatical aspects which the entry of a certain term doesn't clarify, etc. For example, strategies to learn chinese characters and their respective pronunciation (homophones), or arabic broken plurals (unpredictability). --Backinstadiums (talk) 06:02, 2 June 2017 (UTC)Reply[reply]

I'm very much interested in having a permanent discussion around this. You're interested in learning strategies in general, or the role of Wiktionary? I think a more general learning discussion might be a bit off-topic here (there are already plenty of resources on the web), I'm more interested is how Wiktionary can be part of the learning process. I recently ran a workshop “How to learn a language by building a dictionary”. Making / editing entries forces you to think about the language in a structured and formal way, plus you can include your own material (photos you have taken, example sentences you came across, citations from books you read etc.). There was interest in it but the problem is (again and again) that Wiktionary is still unknown and that contributing is difficult. – Jberkel (talk) 07:32, 17 July 2017 (UTC)Reply[reply]
I would think that it would make sense to have a project page in the Wiktionary namespace which interested users could put on their watchlists. Any issues that emerge that would benefit from a wider audience of Wiktionarians could be brought to WT:BP. DCDuring (talk) 12:25, 17 July 2017 (UTC)Reply[reply]

Parameters of Template:quote vs Template:quote-book etc.[edit]

Currently we have {{quote}}, which is just for formatting the text of a quote much like {{usex}}. Then we have {{quote-book}} and its relatives, which also show the source info above the quote, and are a lot more elaborate and contain many parameters. {{quote}} is quite easy to use, since it works the exact same way as {{usex}} and people are very familiar with that. I think it would be desirable if the other set of templates could be modified so that they are more compatible with {{quote}}, so that the more elaborate templates are easier to use for people (like me) who are used to {{quote}} and {{usex}}. I'm thinking mainly of the parameters: first parameter is language, second is quote, third is translation. Would this be ok? —CodeCat 18:56, 1 June 2017 (UTC)Reply[reply]

I'm OK with that. You might have to do a lot of cleanup though, since language codes aren't required in {{quote-book}} right now, even if there is a translation. DTLHS (talk) 18:58, 1 June 2017 (UTC)Reply[reply]
Yes, and I believe it accepts language names rather than language codes too. This is a drawback of these quote templates currently: they don't tag the quoted text, which {{quote}} does. —CodeCat 19:01, 1 June 2017 (UTC)Reply[reply]
Pinging @Smuconlaw as a person who has worked a lot on these templates. —CodeCat 20:37, 1 June 2017 (UTC)Reply[reply]
I'm not very clear as to what is being suggested. All the {{quote-}} family of templates can be used with positional parameters for the basic parameters; for example, {{quote-book|[year]|[author]|[title]|[url]|[page]|[passage]}}. I don't understand what "tagging the quoted text" entails, nor why adding a language code as the first parameter is needed. The {{quote-}} (and {{cite-}}) templates have not hitherto had a language code parameter, so if such a parameter is added a bot would need to update all the uses of the template. — SMUconlaw (talk) 20:29, 2 June 2017 (UTC)Reply[reply]
All templates on Wiktionary that display words in another language will wrap text in a bit of HTML that indicates the language of the text, by convention using either the first parameter to indicate the language, or the lang= parameter, depending on the template. However, this is missing for the quote- templates. They have a language parameter named language=, but it's optional, and it's provided with a language name rather than a language code. It doesn't follow the conventions of other templates. What I would like is if the first three parameters of the quote- templates could be the same as those of {{quote}}, with the others being named parameters: {{quote-book|[language code]|[passage]|[translation]|author=[author]|page=[page]|title=[title]|url=[url]|year=[year]}}. The current transliteration= parameter would be renamed to tr=, again to match other templates. —CodeCat 20:36, 2 June 2017 (UTC)Reply[reply]
I guess I have no strong objection to changing the first three positional parameters so long as a bot can come along and carry out all the required changes to pages where the templates have been used. However, I'm afraid I have no experience with adding language tagging to templates. What is such tagging for, actually, and which text is tagged – the quotation? — SMUconlaw (talk) 21:11, 2 June 2017 (UTC)Reply[reply]
Script tagging goes along with language tagging, allows the text to be displayed in appropriate fonts, which are specified in MediaWiki:Common.css. This is particularly important for non-Latin scripts, or Latin scripts containing unusual characters. Language tagging (I hear) tells screen readers which language the text is written in, allowing them to read it correctly. Script tagging is generally added at the same time as language tagging is added, by Module:script utilities. Linking templates like {{l}} and {{m}} add script and language tagging using this module, as well as {{usex}} and {{lang}}. — Eru·tuon 22:00, 2 June 2017 (UTC)Reply[reply]
If you want to see it in action, go to Special:ExpandTemplates and type in {{quote|en|this is a quote}}. —CodeCat 22:04, 2 June 2017 (UTC)Reply[reply]

A category for words like legged and learned[edit]

I’d like us to have a category for words with -ed pronounced a full -èd where -’d is expected (in other words, it wouldn’t include words like pitted and modded), but I can’t think of a good, accurate name. Any ideas? — Ungoliant (falai) 20:28, 1 June 2017 (UTC)Reply[reply]

Words with unexpected syllabic -ed? — Eru·tuon 20:37, 1 June 2017 (UTC)Reply[reply]
Are you talking about homographs with different standard pronunciations and distinct meanings, but not distinct etymologies? Some more: aged, cussed, dogged. Would you include words that just had different (presumably standard) pronunciation of the -ed like alleged. DCDuring (talk) 22:58, 1 June 2017 (UTC)Reply[reply]
Why not something like Category:English heteronyms (-ed). One could see how to generalize the category name to other morphemes or morpheme groups, though the scheme would probably not work for all heteronyms in all languages. DCDuring (talk) 23:02, 1 June 2017 (UTC)Reply[reply]
Hmm, heteronym apparently means a word with a different pronunciation and meaning. The category we're speaking of should only relate to pronunciation. — Eru·tuon 23:33, 1 June 2017 (UTC)Reply[reply]
legged and learned have common etymologies and related meanings, but not identical meanings. alleged is, I think the only one of those so far mentioned with the same meaning for both pronunciations. DCDuring (talk) 23:49, 1 June 2017 (UTC)Reply[reply]
I assumed from the original post that the category would only relate to pronunciation, because @Ungoliant MMDCCLXIV didn't mention meaning at all. — Eru·tuon 00:31, 2 June 2017 (UTC)Reply[reply]
Yes, I’m talking about a category for words that have -ed but don’t follow the suffix’s usual pronunciation rule (as Erutuon explains below), regardless of the relationship between its senses and regardless of whether the same word is pronounceable in a way that follows the rule. — Ungoliant (falai) 01:30, 2 June 2017 (UTC)Reply[reply]
For more context, the rule is that the suffix -ed is pronounced /d/ or /t/ after most consonants, but /ɪd/ or /əd/ after /d/ or /t/. (The difference between /ɪd/ and /əd/ is dialectal; some dialects have one, some have the other.) So the category would be for words in -ed that don't follow this rule, that should have /d/ or /t/, but have /ɪd/ or /əd/ instead. — Eru·tuon 00:31, 2 June 2017 (UTC)Reply[reply]

Here's a comprehensive summary for -edness, -edly . --Backinstadiums (talk) 05:58, 2 June 2017 (UTC)Reply[reply]

Category:English terms with unexpected syllabic -ed it is. — Ungoliant (falai) 20:57, 7 June 2017 (UTC)Reply[reply]
I don't like the word the "unexpected". Is there a better way to exclude -ted and -ded? --WikiTiki89 21:30, 7 June 2017 (UTC)Reply[reply]
As a native speaker I certainly expect them or at least allow for their possibility, except for the ones based on misspellings (wretch) or with one of the pronunciations being for a rare word, usually a verb (eg, sacre).
Do I understand correctly that the motivation for the category is that someone who encountered in writing the word being used in a sense which commonly required the separate syllable for -ed would mispronounce it, having insufficient experience hearing it? If so, this seems a much better Appendix than a category. A usage note containing a link to such appendix would be more useful than a category. I think an appendix would allow for more flexibility (and length) in its title. DCDuring (talk) 00:34, 8 June 2017 (UTC)Reply[reply]
I think the usefulness is simply that someone may be interested in seeing a list of words in this category, which is exactly what categories are for. --WikiTiki89 15:27, 8 June 2017 (UTC)Reply[reply]
@Wikitiki89 feel free to rename it to anything you think will be accepted. — Ungoliant (falai) 15:23, 8 June 2017 (UTC)Reply[reply]

I think we're missing this. For example, зачем (začem) is not a prefixation, it's really the preposition за (za) with its instrumental regimen чем (čem) (compare French pourquoi). --Barytonesis (talk) 12:05, 2 June 2017 (UTC)Reply[reply]

Wouldn't that fall under {{compound}}? -- GianWiki (talk) 11:07, 16 June 2017 (UTC)Reply[reply]
Sounds like it would be a subtype of compound: one in which the elements were formerly separate words in a phrase. — Eru·tuon 18:23, 16 June 2017 (UTC)Reply[reply]
Sounds like a great idea. It would be useful in the Greek etymon of ephemeral, for instance. — Eru·tuon 18:23, 16 June 2017 (UTC)Reply[reply]
@Erutuon: I went ahead and created Category:Univerbations by language as well as Category:Russian univerbations. I'd like to create {{univerbation}}, but my ineptness at computing prevents me from doing anything other than copy-pasting the code of other templates; yet I don't want to copy-paste {{back-formation}} or {{doublet}} or {{unknown}} and so on, because I understand they're all using obsolete parameters like "lang", and will eventually be updated. --Barytonesis (talk) 19:43, 27 June 2017 (UTC)Reply[reply]
@Barytonesis: I was going to make {{univerbation}}, but I'm not quite sure what parameters it would need. For instance, how would you like to encode the example above, зачем (začem)? Something like {{univerbation|ru|за чем}}, or {{univerbation|ru|за|чем}}? — Eru·tuon 04:27, 29 June 2017 (UTC)Reply[reply]
@Erutuon: Good question. Several thoughts (I will depart a little from your question, sorry about that):
  • Do we plan to eventually create categories for compounds like Category:English terms derived from "apple", as it was suggested here? If so, then I guess we could do the same for univerbations (and in that case, the code would have to be {{univerbation|ru|за|чем}}, wouldn't it?). But that idea strikes me as inappropriate in the case of univerbations (Category:Russian univerbated syntagms containing за doesn't make much sense); and makes me actually think that univerbations are rather a sister category to compounds than a subtype of them.
  • In any case, I think that no "+" sign should appear in the output: both markups should give "за чем" rather than "за + чем".
  • For these two reasons, right now I don't really feel strongly one way or the other.
  • As I mentioned in this vote, I'd rather have no automated text at all in etymology templates, and type "Univerbation of" in plain text. What is your opinion about that? --Barytonesis (talk) 10:17, 29 June 2017 (UTC)Reply[reply]
If we did want to add categories for univerbations containing individual words, it would be possible to do that with either {{univerbation|ru|за чем}} or {{univerbation|ru|за|чем}}, but I think {{univerbation|ru|за|чем}} would make it easier. Also, {{univerbation|ru|за чем}} is more prone to human error: it would be easy for an editor to fail to link the words. I suppose {{univerbation|ru|за чем}} is another option; it could be linked in the same way that {{head|en|phrase}} adds links for each word automatically.
I agree about having no plus sign between the constituent words, since they form a natural phrase.
I was going to say that if we use {{univerbation|ru|за|чем}}, there might be the annoyance of having separate |altN= parameters for each word, to add accent marks to Russian words and macrons to Latin words (and so on). For instance, the univerbation ἐφήμερος (ephḗmeros) would have to be encoded as {{univerbation|grc|ἐπί|alt1=ἐπῐ́|ἡμέρα|alt2=ἡμέρᾱ}}. But actually, it could just be written as {{univerbation|grc|ἐπῐ́|ἡμέρᾱ}}, and the linking module would automatically generate the actual entry titles.
I'm sympathetic to not having manual text "univerbation", but it might be a good idea to link to a definition in Appendix:Glossary, since few people are likely to know the term, so I'm not sure. — Eru·tuon 18:45, 1 July 2017 (UTC)Reply[reply]

Proposal: Clean up, rename and replace "en:" → "English" in all categories[edit]

(This obviously would need a vote to be implemented.)

In the past, some old categories like Category:es:Japanese derivations and Category:es:Derogatory were renamed to remove the language code. I think this was an improvement. (related votes: Derivations categories, Lexical categories)

Please check if the grammar is OK everywhere. Feel free to make any corrections, suggest any changes or ask any questions.

See also discussion: User talk:-sche#Properly splitting topic and set categories.

Place names (see also WT:Place names for naming conventions, edit that page if needed)

Proposal: ALWAYS add the country at the end when applicable.

English ... jargon

(to avoid questions like: is "Medicine" for medicine jargon or for terms relating to the medicine?)

English names of ... (for proper nouns?)
English terms relating to ... (or "pertaining to", or "involving", etc.)

Reason for "relating to" -- most or many of the description of these categories use "related to"

--Daniel Carrero (talk) 14:16, 2 June 2017 (UTC)Reply[reply]

It seems somewhat ok, hesitantly. My biggest gripe at the moment is that it's not Category:English names of cities in Ontario, Canada. The addition of country names is a definite improvement. I'll abstain for the moment on the "relating to" categories. —CodeCat 14:43, 2 June 2017 (UTC)Reply[reply]
Oops! I made a mistake and typed Category:English cities in Ontario, Canada without "names of". You made me realize that, and I fixed it in the list above. --Daniel Carrero (talk) 14:45, 2 June 2017 (UTC)Reply[reply]
I don't know that I disagree as such but I think this will cause a lot more problems than it solves and it seems like a huge undertaking for very little benefit. —Justin (koavf)TCM 15:53, 2 June 2017 (UTC)Reply[reply]
I see. What problems do you think it will cause? FWIW, see Wiktionary:Votes/2017-03/Request categories 2 for another large category renaming project that was voted and approved, and was successfully implemented. --Daniel Carrero (talk) 15:58, 2 June 2017 (UTC)Reply[reply]
@Daniel Carrero: The wording of many of these categories. "English terms referring to [X]" or "English terms related to [X]"? "English words for [Y]" or "English language words for [Y]", etc. As it stands now, the scheme is very straight forward: code:Idea. —Justin (koavf)TCM 16:32, 2 June 2017 (UTC)Reply[reply]
The current scheme is also ambiguous about whether a category is for words in a given set, or words related to a topic. This has been causing quite a few headaches lately. Is Category:Stars for names of individual stars, words for types of stars, or any words related to stars? What if we want categories for each of these types? Daniel's scheme at least avoids the ambiguity: Category:English names of stars, Category:English terms related to stars. I'm not sure where words for types of stars would go. —CodeCat 16:36, 2 June 2017 (UTC)Reply[reply]
Shall we introduce a new type of category to the proposed list? I'm thinking of this:
There's a 2011 vote that concerned what to do with a few categories, including specifically Category:en:Stars, but the approved rule is not being followed right now. According to Wiktionary:Votes/2011-08/Categories of names 2, Category:en:Stars must contain only names of stars, and never other star-related terms like fixed star, quadruple star, quadruple star system, etc. The category description says the same thing. But the name Category:en:Stars does not help, it could easily contain any of those "forbidden" terms.
Another example of a currently confusing category name: In practice, Category:en:Internet contains two things: terms involving the internet (frame, FTP, e-mail...) and terms used in the internet (FWIW, IOW, IYKWIM...). The latter should actually be in Category:English internet slang, but once again the category name doesn't help -- "en:Internet" could mean either of the aforementioned possibilities. If this proposal passes, Category:en:Internet should be renamed to Category:English terms relating to the internet (I guess the "the" fits here, right?). Feel free to discuss about different wordings, other than "relating to", but I believe we can always say just "English", never "English language" -- after all, we use Category:English nouns, never Category:English language nouns.
Aside from that, the place name categories need to be cleaned up one way or another, naturally. --Daniel Carrero (talk) 16:53, 2 June 2017 (UTC)Reply[reply]
What's the supposed benefit of using language names rather than language codes? Names can be awkwardly long, and are not guaranteed to be unique. Equinox 19:59, 2 June 2017 (UTC)Reply[reply]
They are guaranteed to be unique, per WT:LANG. And the benefit is that users can't be expected to learn language codes merely to use Wiktionary. It's an unnecessary barrier. —CodeCat 20:00, 2 June 2017 (UTC)Reply[reply]
Yes, they are guaranteed to be as unique in category names as they are in L2 headers. Which means, there could be cases where people assume "Riang" means the Bangladesh language ria when it actually means the Burma language ril, but that shouldn't cause any more headaches with categories than it does with L2 headers. - -sche (discuss) 22:50, 2 June 2017 (UTC)Reply[reply]
There are two components to this proposal: changing language code to language name, and changing the names of the categories (the part after the language code or name). Hypothetically the second could be done without the other: that is, there could be a chimeric category name, Category:en:Names of cities in Kyoto Prefecture, Honshu, Japan. I suppose it makes sense to do both at the same time, since (I imagine) thousands of categories will have to be renamed when either change is made. However, some editors might want to keep language codes but have the other part of the category name be changed. (The other option, using language names but not changing the rest of the category name, will not work: Category:en:SexCategory:English sex is nonsensical.)
On the whole, the category names are clearer, but they will be more difficult to find when one is categorizing, because they are longer. Admittedly, the existing categories are difficult to find too, especially when creating one that doesn't exist for a particular language yet. I wonder if a tool could be created to make this easier. HotCat isn't quite what I'm thinking of. Perhaps something where you could type in a phrase, like "cities kyoto" and find Category:English names of cities in Kyoto Prefecture, Honshu, Japan, or the umbrella category thereof.
Functionally, the only purpose of using language codes rather than names is to distinguish topic or set categories from the categories that use language names (part-of-speech categories especially). If language names are used, the only distinction will be the category names after the language name. So, the two types of categories will be harder to tell apart. — Eru·tuon 21:48, 2 June 2017 (UTC)Reply[reply]
Category:English sex is nonsensical, but the colon doesn't have to be deleted, so Category:English:Sex. —CodeCat 22:00, 2 June 2017 (UTC)Reply[reply]
I would indeed favour something like that over "sentence" categories like "English terms involving/about/relating to sex". The latter seem a bit wordy and would probably sound actively bad or strange in certain cases. BTW, regarding Daniel's proposal, I think "jargon" is a very poor choice: it is a loaded term, suggesting that this is needlessly complex language; we did in fact remove "jargon" glosses from a lot of entries at one stage. Equinox 22:21, 2 June 2017 (UTC)Reply[reply]
I have similar concerns/dislike as Erutuon and Equinox towards the long names. "English names of cities in Kyoto Prefecture, Honshu, Japan" is horribly long, and probably too finely "granular" (but the latter is a separate issue). Long names are harder for users to write when adding or searching for categories.
Language codes have several benefits over language names, including being shorter and not needing to be moved when we rename a language (moving 100 categories is a major hassle which I suppose we can avoid now by mass-deleting the categories and letting bots create the new cats and WikiData sort the interwikis out). OTOH, I understand those who feel they are a barrier for less-adept users.
Can we avoid the unwieldy "sentence" names and make the topic category "English:Foobar", and the set/list category "English:List of foobars"? (Another idea, proposed on my talk page, is "Category:English:topic:Foobar" vs "Category:English:List:Foobar(s)".)
As others have said, we should avoid calling things "jargon". Maybe "terminology" would work. - -sche (discuss) 22:42, 2 June 2017 (UTC)Reply[reply]
Re: "Functionally, the only purpose of using language codes rather than names is to distinguish topic or set categories from the categories that use language names (part-of-speech categories especially)." -- It's not been always like this in the past, like in the old categories mentioned in the op (Category:es:Japanese derivations and Category:es:Derogatory). Even if that rule is supposed to be followed now, we have Category:English female given names, Category:English surnames, etc., and if we rename any category to contain "names", "terms", "jargon" or "terminology" in the name, it technically becomes a "lexical" category and will need to start with "English..." as per this rule.
Are codes better than names always, or are names better than codes always? If we had no categories for derivations whatsoever, would we want to create Category:es:Japanese derivations? The proposed category names like Category:English terms related to sex are supposed to be straightforward -- the category contains what the title says, as in a normal English text.
The granularity of place name categories is optional. I proposed Category:English names of cities in Kyoto Prefecture, Honshu, Japan, but it could be Category:English names of cities in Japan, even though I prefer the longer name.
By the way, if we renamed all categories as proposed above we could remove "by language" from all categories. In the current naming scheme, we have this:
If this proposal passes, we could have this, without "by language" anywhere (unless we want to keep the "by language", of course, but that won't be a requirement like today):
Let's compare this: Category:English medicine terminology and Category:English medicine jargon. Is "terminology" really better? I wouldn't like the category to have common terms like disease, heal, doctor, check-up. So I feel that "jargon" is better for the purpose of avoiding these terms.
Additional proposal: we could have one module listing all catgegories with "names of", another module for "terms for", another for "terms relating to", and another for the list of place names. --Daniel Carrero (talk) 02:16, 3 June 2017 (UTC)Reply[reply]

We seem to have many different ideas. Let's have a poll to get a rough tally of whether more people like names vs codes, and whether more people like shorter/condensed or longer/descriptive names. That way, we can see what kind of names we should focus on (like, if most people want descriptive names, then we can focus on deciding whether "pertaining to" or "involving" is better, but if most people want short names, then bikeshedding the format of hypothetical long names would be silly). - -sche (discuss) 22:42, 3 June 2017 (UTC)Reply[reply]

Categories for terms which are limited to the "jargon" of medicine (if you are distinguishing them from "terms that pertain to the topic of medicine") are neither "topic" nor exactly "list" categories, IMO, but are in the same vein as categories for terms which are limited to British English. So, I don't think my poll covers them, because I think they should be addressed separately. - -sche (discuss) 22:42, 3 June 2017 (UTC)Reply[reply]

There are way too many reasons not to like this, so I'll start with just one: all of these ideas move the interesting information farther and farther to the end of the name "blahblahblahblahblahblahblah fish" in newspaper jargon terminology talk this is known as burying the lede. The way sorting works means we can't rearrange the order, so we need to be as concise as possible in the left-hand parts. Take a look at the massive logjam of categories at the bottom of most multilingual entries. Now imagine it being twice the size without a corresponding increase in the size of the rightmost nodes. Someone mentioned HotCat: the auto-complete feature is going to be pretty much useless if you have to type in pretty much the whole category name before it gets to the part that it can fill in. And all the stuff about descriptive English sentences being easier for people to understand was the main selling point of COBOL- remember COBOL? I'm sure every programmer aspires to someday write code like the immortal "ADD A TO B GIVING C", and I'm sure any kid in grade school could debug three COBOL business applications before breakfast... right? >;p Chuck Entz (talk) 09:07, 4 June 2017 (UTC)Reply[reply]
@Chuck Entz: Do you think you could at least agree with me that the current language-code categories need to be cleaned up one way or the other? I'm pretty sure we can't say that the current system is acceptable. One of Wikimedia's values is "We strive for excellence." Place name categories are probably the messiest of all.
I don't mean it in a sarcastic way, it's just a normal question: As you know, I prefer language names instead of codes and I gave my reasons in this discussion. But if language codes were the best option always, shouldn't all language-specific categories like Category:English nouns and Category:Italian terms derived from Ancient Greek be renamed to include language codes instead of names? When we use the auto-complete feature, I believe we have to type at least "Italian terms d" if we want to get Italian terms derived from other languages. Or we just navigate to Category:Italian terms derived from other languages (which is named using normal English text, like the other categories I'm proposing).
Based on this discussion and this vote, I think in 2011 I myself helped to introduce and cement our existing tradition that apparently "lexical" categories have language names and "topical" categories have language codes, but it was just because I wanted to clean up some categories and put forward the proposal of replacing say Category:es:Euphemisms into Category:Spanish euphemisms without necessarily having to change the whole system yet.
I don't think the logic of programming languages applies to our categories. COBOL's "ADD A TO B GIVING C" is not better than "A+B=C", but Category:English names of stars has some merits discussed elsewhere in this discussion, as opposed to Category:en:Stars. If we want a category for medicine slang (multileaf collimator? s/p? CINV?) which cannot contain everyday words like heal, disease and doctor, then maybe the word "jargon" (or slang, or terminology) somehow needs to be in the category name.
Just for the sake of discussion, if having short categories with the main idea first is important, I wonder if there's some merit to using names like Category:Medicine (English, jargon), Category:Stars (English, star names) or Category:Stars (English, related terms). I'm not saying I support having these names, I'm just discussing what their merits are. We could also talk about renaming Category:Italian terms derived from Ancient Greek to Category:grc→it and Category:Italian terms derived from other languages to Category:*→it, which is probably one of the shortest options available (not to mention these are shorthands that feel kind of related to programming language logic) where they could still be understood as derivation categories. Again, I'm not trying to be sarcastic. Obviously, I don't think the shortest names are always the better ones. I know "grc→it" is a bad category name (in my opinion, at least), but I'm fine with discussing multiple possibilities. --Daniel Carrero (talk) 10:37, 4 June 2017 (UTC)Reply[reply]

Another category where this is an issue: Category:en:Forests. While the description and parent category suggest it's for names of forests, it also contains terms for types of forests, and terms related to forests. —CodeCat 22:10, 28 June 2017 (UTC)Reply[reply]

Poll 1: Language names vs codes[edit]

On a new line, please indicate what you support:

  • (1) language names (like "French") to be used in naming topic categories (like categories for terms pertaining to the topic of religion);
  • (2) language names to be used in set/list categories (like lists of dog breeds); or
  • (3) language codes (like "fr") to be used in naming topic categories;
  • (4) language codes to be used in set categories.
Thus, if you want names to be used in all cases, you can indicate "1&2", or if you want codes in all cases, then indicate "3&4". But if you want a mix like "1&4", you can indicate that.
  1. Abstain for now; both codes and names have both benefits and drawbacks. - -sche (discuss) 22:42, 3 June 2017 (UTC)Reply[reply]
  2. 1&2 —CodeCat 22:55, 3 June 2017 (UTC)Reply[reply]
  3. Support 1&2.
    As I proposed above and then further commented in my reply to "Poll 2" below, I'd prefer category names that read like normal English, like Category:English terms relating to sex. In "real life", are many people even aware of different ISO codes for each language, not to mention our non-ISO made-up codes like "alv-pro", "nai-dly" and others found in Module:languages/datax?
    Currently, Category:Baseball contains 18 subcategories like "en:Baseball", "fr:Baseball", "ko:Baseball", etc. Sure, we all know what they mean, but navigating them requires learning how ISO handles codes and how we separately handle them. Language codes are gibberish if you don't know what they mean. — Sure, even "I like monkeys." is gibberish if you don't know its meaning, but you get the point. Using normal text with the actual language name should make the category contents immediately obvious to English speakers.
    Categories starting with our specific set of ISO and made-up language codes are an English Wiktionary signature, and therefore a barrier for reuse. Anyone is allowed to copy Wiktionary content - they can create mirrors, books, CDs with it. So, it's better to make the material as "generic" as possible. If a reader is navigating a new site called www.definitelynotwiktionary.com and finds a category name like Category:gmq-bot:Music, they will probably not understand what "gmq-bot" means. Even if we generously assume that the reader has the privilege of being knowledgeable about our language codes, then it's possible they will be compelled to think "oh so we're using that English Wiktionary system now, I'd better get my list of language codes that they use to start navigating categories like this". By contrast, if we renamed Category:gmq-bot:Music to Category:Westrobothnian terms relating to music, the new name should be sensible anywhere. --Daniel Carrero (talk) 05:07, 4 June 2017 (UTC)Reply[reply]

Poll 2: Longer vs shorter names for set categories[edit]

On a new line, please indicate what basic type of name you support — for the part of the category that comes after the language name or code:

  • (1) Long descriptive names like "names of municipalities in São Paulo, Brazil" and "names of dog breeds" or "terms for dog breeds". (The precise format, like "names of" vs "terms for", can be worked out next if it's clear that people prefer long descriptive names to short names.)
  • (2) Short names that contain "list" or "set" to distinguish list vs topic categories: like "list:municipalities in São Paulo, Brazil" and "list:Dog breeds" or "set:Dog breeds".
  • (3) Very short names that don't distinguish list vs topic categories: like "municipalities in São Paulo, Brazil" and "Dog breeds".
  1. Prefer (2). Oppose (3) as too ambiguous. - -sche (discuss) 22:42, 3 June 2017 (UTC)Reply[reply]
  2. I believe (1) is the best option, because names like Category:English terms relating to sex are to be read as normal English text, like other categories already are. Oppose (2) because the distinction between list, or set, or topic is arbitrary and not immediately obvious. And of course, oppose (3) as too ambiguous. See further comments below.
    If you get all terms relating to Christianity, this is equally a list, a set and the topic of Christianity-related terms. If we choose to pretend these words have unique meanings which clearly set them apart (which they don't), this would be an obvious kludge to avoid category overlap using as few characters as possible. Using "list", "set:", "topic:" at the start of categories is like using obscure abbreviations to save a few characters. (Some people oppose using template abbreviations like {{der}} instead of {{derived}}, saying that the latter is easier to read. What to do with templates is a separate discussion, but that is a good point. Aside from that, I believe we have a consensus not to use abbreviations like q.v., L., Gr., esp., cf., &c. in etymologies.) If we implemented these arbitrary category prefixes, I fear we would probably have to constantly lecture anons and new editors about what is the correct category prefix to use.
    What I'm saying here is consistent with what I proposed above. The proposed category names are supposed to be read as normal English text. Currently, we have to wonder if categories like Category:en:Stars contains names of stars, and/or types of stars, and/or terms relating to stars. If this proposal passes, we may have this:
    I'm not saying all the three "star" categories above will have to be created, but I'd like any existing category to conform to a naming system like this. The wording may change if people want, naturally.
    In the past, other proposals that made category names read like normal English text were voted and approved:
    --Daniel Carrero (talk) 04:25, 4 June 2017 (UTC)Reply[reply]
  3. Prefer (1). Category names should describe exactly what they contain. DTLHS (talk) 05:11, 4 June 2017 (UTC)Reply[reply]
  4. 1&2. Oppose 3 as too ambiguous as well. —CodeCat 15:31, 4 June 2017 (UTC)Reply[reply]

Poll 3: Longer vs shorter names for topic categories[edit]

On a new line, please indicate what basic type of name you support — for the part of the category that comes after the language name or code:

  • (1) Long descriptive names like "terms pertaining to Christianity" or "terms relating to Christianity". (The precise format can be worked out next if it's clear people prefer long descriptive names to short names.)
  • (2) Short names that contain "topic" to distinguish list vs topic categories: like "topic:Christianity".
  • (3) Very short names that don't distinguish list vs topic categories: like "Christianity".
  1. Prefer (2). Oppose (1) as too long (cf Chuck's comments further up this thread), and oppose (3) as too ambiguous. - -sche (discuss) 22:42, 3 June 2017 (UTC)Reply[reply]
  2. I believe (1) is the best option, because names like Category:English terms relating to sex are to be read as normal English text, like other categories already are. Oppose (2) because the distinction between list, or set, or topic is arbitrary and not immediately obvious. And of course, oppose (3) as too ambiguous. See further comments in my response to the "Poll 2" above. --Daniel Carrero (talk) 04:25, 4 June 2017 (UTC)Reply[reply]
  3. (1) DTLHS (talk) 05:12, 4 June 2017 (UTC)Reply[reply]
  4. 3: Category titles are supposed to be as short as possible. Purplebackpack89 15:30, 4 June 2017 (UTC)Reply[reply]
  5. 1&2. Oppose 3 as too ambiguous as well. —CodeCat 15:31, 4 June 2017 (UTC)Reply[reply]
  6. Either 1 or, preferably 2, because they permit the addition of usage-context categories (eg, military slang for woman, member of the local population, etc.). DCDuring (talk) 19:51, 1 July 2017 (UTC)Reply[reply]


From the results of the poll above, I gather that people want language names in the categories, but that there is less consensus on the format of the rest of the name. What happens next? —CodeCat 19:26, 1 July 2017 (UTC)Reply[reply]

Sorry, I'm being lazy (or "delegating"; there's other stuff I want to focus on). This should probably be a prefix though: we have enviro-friendly, environazi, environut, Enviropig, envirospeak, envirotard. Most currently have blend etymologies. Equinox 00:57, 3 June 2017 (UTC)Reply[reply]

Done. —Vorziblix (talk) 03:23, 4 June 2017 (UTC)Reply[reply]

June Lexisession: concert[edit]

A concert.

Monthly suggested collective task is to take care of concert. You already have a Wikisaurus:musical instrument and a Wikisaurus:musical composition but nothing about the show! Well, in June, there is the Fête de la Musique [World Music Day] and that's enough to plan a new Wikisaurus:concert, isn't it? Also, there is plenty pictures on Commons to illustrated entries related to musical performances.

Show must go on!

By the way, Lexisession is a collaborative experiment without any guide nor direction. You're free to participate as you like and to suggest next month topic. If you do something this month, please report it here, to let people know you are involve in a way or another. I hope there will be some people interested by playing music Noé 09:21, 3 June 2017 (UTC)Reply[reply]

Where to place {{wikipedia}} templates?[edit]

@Atitarev, Cinemantique, Wikitiki89, CodeCat In new entries, I've been putting {{wikipedia|lang=ru}} templates just under the ==Russian== headword, but some existing entries put in under the ===Noun=== or similar headword. See атеи́зм (atɛízm) for an example where I moved it up. Not sure if this is correct, comments? Also, is there a difference for English-language and foreign-language Wikipedia references? An example where I put both is пого́ст (pogóst); this Russian term has several meanings specific to Russian culture and has an English-language entry under pogost, which is helpful in explaining some of the meanings. Benwing2 (talk) 20:50, 3 June 2017 (UTC)Reply[reply]

For English terms I used to place them under the ==English== header too. However, another editor tended to remove and replace them with:
===Further reading===
* {{pedia}}
so that's what I do now. — SMUconlaw (talk) 20:59, 3 June 2017 (UTC)Reply[reply]
IMO {{wikipedia}} should go under the language header; I find it a bit messy if it's under the POS, maybe unless it only applies to e.g. the noun section of an entry that lists a verb first, in which case I understand putting it underneath the relevant headword line template. Alternatively using {{pedia}} as Smuconlaw describes is also fine. - -sche (discuss) 01:01, 4 June 2017 (UTC)Reply[reply]
The situations with Translingual (taxonomic) is very different than for other languages: the pedia, species, and commons links are on all fours with other external links (which can be numerous). Taxonomic entries often have images so the right hand side can become cluttered and push into other language sections. Thus it makes more sense to me to put the sister-project links under "References".
Something similar applies to entries for English vernacular names of organisms, which often have either images or many sister-project links.
Finally I have the ToC on the right hand side, which further pushes right-hand side content down into other language sections.
I don't know how applicable this is to the use of such links in other L2s. DCDuring (talk) 05:04, 4 June 2017 (UTC)Reply[reply]
Thanks for all the comments. Benwing2 (talk) 23:41, 11 June 2017 (UTC)Reply[reply]
I'm one of the editors who tends to convert {{wikipedia}} to {{pedia}} in Further Reading. The main reason is consistency: why should Wikipedia links get a special treatment? The box is also quite big and takes up a a lot space on the screen, especially when multiple links are stacked. Even the {{wikipedia}} documentation page says: "Consider instead using the inline version of this template". – Jberkel (talk) 06:15, 15 June 2017 (UTC)Reply[reply]
I prefer all sister-project-link templates be placed immediately under the language heading. - [The]DaveRoss 13:35, 15 June 2017 (UTC)Reply[reply]
Hmm, there's conflicting documentation and practice. Should we have a vote on this? – Jberkel (talk) 21:53, 15 June 2017 (UTC)Reply[reply]

What are / were / should be the rules for anagrams in English and other languages?[edit]

I would be willing to run a bot to update anagram sections (in which subset of languages?) if I know the rules. "Rules" meaning, which characters to ignore / normalize, minimum length, etc. DTLHS (talk) 01:26, 4 June 2017 (UTC)Reply[reply]

  • I assume that the rules are the same for usage of the {{also}} template (that also needs bottifying). SemperBlotto (talk) 05:57, 4 June 2017 (UTC)Reply[reply]
It would be great if we created a page which documented character normalization policies in a machine-readable format (Wiktionary:Character normalization?). I know these exist to some extent in user-space, but a unified version would be better. - [The]DaveRoss 12:45, 29 June 2017 (UTC)Reply[reply]

French Wiktionary monthly news - Actualités[edit]


I am dazzled to inform you that the 26th issue of Wiktionary Actualités just came out in English!

As usual, Actualités is in English but talk about French Wiktionary and lexicography in general.

This time: a focus on proposals for Wikimania 2017 related to Wiktionaries, a presentation of two dictionaries about the body and some words about Guaraní language. There is also a stack of statistics, shorts and a game!

As usual, it is translated in English by non-native speakers, so it is not perfect, but can be improved by readers (wiki-spirit and all). Please note that we do not received any money for this publication and we translate it because we are eager to read the same kind of publication about your project in the future and be inspire by your projects. Feel free to leave us comments! Noé 10:12, 4 June 2017 (UTC)Reply[reply]

Fascinating! Hmm, what is en.Wikt doing that could be reported in such a publication?
Last time I recall us comparing the new words other dictionaries had added to our entries, we similarly already had most of them.
We've been working on templates/scripts that would enable an entry to specify its most recent etymon, and have the script find and display that term's etymon, and that etymon's etymon, to reduce content duplication and dissynchronization.
Efforts to create a module that can automatically transliterate vocalized Hebrew are continuing and may lead to a proposal to Unicode to encode separate codepoints for big and small shvas like big and small qamats.
We've been expanding our coverage of languages that (are not dialects of other languages and) do not have ISO codes (Module:languages/datax).
We've been expanding the number of languages we have referenced/verifiable entries in, using in some cases fr.Wikt's entries (which in many cases were based on en.Wikt's redlinked translations of water, ha).
- -sche (discuss) 02:19, 6 June 2017 (UTC)Reply[reply]
Sure, we can report hot topics of en.wikt! Easily if someone from the project do a summary, just like you did. We tried in some old issues, but it was biased because discussions are split in several pages and we didn't know enough the names of the participants to get the whole picture.
The comparison with other dictionaries seems to me very different in English than in French because the editorial choices of French dictionaries is to select only around 60.000 entries whether English dictionary select an average of 100.000. It is purely arbitrary. So in French dictionaries, you will not find any words for technical practices such as leather work for example. Each year, they delete words from dictionaries to save space for the new ones, and I think it is definitively a plus for French Wiktionary, because we do not, and for readers of "old" books, definitions will only be available in Wiktionary, and not anymore in dictionaries! So, French Wiktionary can pretend to have a better coverage than published dictionaries. I think it is more difficult to communicate on this matter for English Wiktionary. Well, I hope you will disconfirm.
Templates for etymologies is a big deal. In French Wiktionary, we prefer to write paragraphs of etymology with large compilations of sources, in a similar way as Wikipedia writing. Policies Wiktionnaire:Étymologie and Wiktionary:Etymology show great differences. French Wiktionary promote long etymologies including folk etymology and false ones with sources. We do not want to have From latin to French in etymology because it is barely false considering the history of the language and the influence of the dialects. We want to trace the path for the forms and the meanings, mentioning regional uses when needed (for example: bataclan). Also, lot of sources says obscure origin when words come from Arab or Gitan (Romani), we prefer to quote these official sources and more recent analyses displaying better data to show that old official sources can be politicly biased. Our etymology have to be more neutral than old ones. I do not judge one strategy better than the other. I think we need to do both. French Wiktionary need to develop a template to display schematic trees with the basic history of words, but also to provide plenty details with the whole history and controversial hypotheses.
Hebrew and Unicode: Great news! I hope you will publish a press release when it will be done!
Expanding coverage for underdescribed languages is great! Is it a global effort or a contribution by few people? In French Wiktionary, it is mainly do by two people, including Pamputt, for water translations Noé 08:44, 7 June 2017 (UTC)Reply[reply]
  • Wiktionary:Milestones and Wiktionary:News for editors are generally used to announce new things. But they're generally not very interesting. I can't imagine many people going "Wow! How cool! Greek nouns have been reclassified from invariable to indeclinable!!!" -WF.

Based on Wiktionary:Beer parlour/2017/May#'character info' box, I created Wiktionary:Votes/2017-06/Allowing character boxes. --Daniel Carrero (talk) 09:55, 5 June 2017 (UTC)Reply[reply]

Language codes for East, South and West Slavic[edit]

I was wondering if it wouldn't be a be a good idea to create language codes for Proto East, South and West Slavic. They're well established and would make sense to reconstruct. --Victar (talk) 23:15, 5 June 2017 (UTC)Reply[reply]

"Proto-East Slavic" is Old East Slavic, is it not? As for West and South, are you sure it's possible to reconstruct as single Proto-West Slavic and a single Proto-South Slavic? --WikiTiki89 23:18, 5 June 2017 (UTC)Reply[reply]
I don't know myself what the differences would be between PES and OES, but if there is no distinction, than I don't think we should have a separate level in descendant trees. I do know that PWS has some pretty distinct features. Matasović argues that South Slavic is strictly as a geographical grouping, not a genetic clade, so I'm not clear on that. @Benwing, Vahagn Petrosyan? --Victar (talk) 01:36, 6 June 2017 (UTC)Reply[reply]
West Slavic is really not that distinct. Most of its features are retentions of things already present in Proto-Slavic. —CodeCat 17:23, 6 June 2017 (UTC)Reply[reply]
Having distinct features does not mean it is possible to have a single consistent reconstruction. And OES is Proto-East Slavic. Don't forget that "Proto" doesn't mean reconstruction, but just that it is the ancestor of a language group. In this case OES is the ancestor of the East Slavic languages, so it is Proto-East Slavic. --WikiTiki89 17:33, 6 June 2017 (UTC)Reply[reply]
I obviously understand what proto means, but if you follow the point I was making, If you're arguing that PES and OES are identical, which I generally disagree with, than we shouldn't have level in a descendant trees for PES above OES, but instead call that branch Old East Slavic. Otherwise that's like adding a Proto Norse above each entry of Old Norse in PGmc descendant trees. --Victar (talk) 18:03, 6 June 2017 (UTC)Reply[reply]
Proto-Norse is a separate language from Old Norse, and happens to be marginally attested, even. At least Old Gutnish does not descend from any attested variety of Old Norse, and I recall hearing about similar archaisms in the Finland Swedish and Jutlandic Danish dialects too. --Tropylium (talk) 23:34, 6 July 2017 (UTC)Reply[reply]
I already do that. I label the line as "East Slavic" but put the OES term on the same line. —CodeCat 18:06, 6 June 2017 (UTC)Reply[reply]
That's great, but I haven't seen that the case in the entries I've come accross, as per my example above. If people are in agreement with that though, I'm satisfied. --Victar (talk) 18:17, 6 June 2017 (UTC)Reply[reply]
First of all, why would it be obvious to me what you personally know or don't know? If I want to make sure we are on the same page in terms of terminology, you shouldn't take it personally. Second of all, OES is the ancestor of all East Slavic languages; that makes it by definition Proto-East Slavic. It's not something you can disagree with unless you want to say that OES is not the ancestor of all East Slavic languages (which you could maybe make a case for regarding the Old Novgorod dialect or North Russian, but in that case we probably can't reconstruct a single Proto-East Slavic anyway). If the tree is wrong, it should be fixed; don't impose incorrect descriptions of reality onto reality itself. --WikiTiki89 18:14, 6 June 2017 (UTC)Reply[reply]

Manichaean Middle Persian[edit]

Manichaean Middle Persian is currently designated as a separate language from Middle Persian, but this isn't the case, as it's simply one of several scripts used. Shouldn't we delete it? @Vahagn Petrosyan? --Victar (talk) 03:29, 6 June 2017 (UTC)Reply[reply]

The difference is not just in the script. Manichaean Middle Persian has systematic dialectal differences from Zoroastrian (Book) Middle Persian. For example, to Zoroastrian nd corresponds Manichaean nn, as in Book bnd (band) : Manichaean bn (bann, bond, link); Old Persian rd gives Zoroastrian l but often r in Manichaean, e.g. sāl vs sār ‘year’; Iranian gives Zoroastrian ar but often ir in Manichaean, e.g. mard vs mird ‘man’.
Even if we decide to merge both under Middle Persian, we should keep Manichaean Middle Persian as an etymology-only language. @ZxxZxxZ, what do you think? --Vahag (talk) 07:13, 6 June 2017 (UTC)Reply[reply]
I believe some of those differences are due to limitations and idiosyncrasies of the scripts itself. Note that even though the transcription BMP s’lyn- (to provoke) contains an l, it is pronounced /sārēn-/, as per Cheung. Both alphabets also lacked a full set of vowels. Even so, I think these are minor dialectal changes. --Victar (talk) 08:00, 6 June 2017 (UTC)Reply[reply]
As Vahag said the differences are beyond the script, not just Manichaean, but also Zoroastrian (Pazend, Avestan alphabet). I've even read there are differences between the Middle Persian written in Inscriptional Pahlavi and the Middle Persian in Book Pahlavi (which was mostly written in Islamic period and is called "late Middle Persian", as opposed to the "early Middle Persian" of the inscriptions), though I'm not aware of any instances beyond spelling differences (e.g. in the arameograms used). Regarding the s’lyn- instance, it's true, though the letter "l" is also used for l, anyway the instance provided by Vahag is a different case: we know ŠNT was pronounced as sāl in Middle Persian, but it is recorded with r in Manichaean alphabet. I think we should keep them separate. --Z 11:30, 6 June 2017 (UTC)Reply[reply]
This family of languages has fluctuated between l and r since the days of IIr. I see more variation in dialects of English than in forms of Middle Persian -- certainly not enough be called a separate language -- and all these "differences" seem highly predictable to me. @-sche, this strikes me as one of those "splittist" cases you mentioned. --Victar (talk) 14:12, 6 June 2017 (UTC)Reply[reply]
On second thought I changed my mind a bit regarding this. --Z 20:37, 6 June 2017 (UTC)Reply[reply]
I am open to being persuaded otherwise by you all, who seem to have greater knowledge of this subject than I do, but based on this discussion and from what Wikipedia says, it does sound like we are dealing with dialects. They should of course have separate etymology codes (like Cajun French vs standard French). If l/r variation also exists within one or especially both varieties and not just as a distinction between them, that would suggest it should not be held up as a reason to separate them; likewise, if the appearance or absence of any particular variation is due to the constraints of script! Wikipedia speaks of using the documents which were written in more expressive/conservative scripts to understand the documents written in the other script. (It makes me think of the ISO granting separate codes to hieroglyphic vs cuneiform Luwian.) - -sche (discuss) 04:56, 8 June 2017 (UTC)Reply[reply]

In this diff, @Metaknowledge reverted my edit to Wiktionary:Criteria for inclusion/Well documented languages.

I'd like to do the same edit again, where I added two vote references. See the history of the page for further comments from him and myself. --Daniel Carrero (talk) 04:00, 6 June 2017 (UTC)Reply[reply]

Addendum: Wiktionary:Votes/2011-04/Sourced policies is the vote where it was accepted to link every piece of text in EL and CFI to their supporting votes through the wiki technique of references. There are too many so-called policies created unilaterally without verifiable consensus. WT:EL and WT:CFI themselves are partially voted and partially non-voted. If we don't link the votes, we can't easily verify the fact Wiktionary:Criteria for inclusion/Well documented languages is thankfully almost 100% voted and approved, with a few unvoted changes concerning Arabic, Irish and Welsh. --Daniel Carrero (talk) 17:22, 6 June 2017 (UTC)Reply[reply]

All words in all languages[edit]

In the very first paragraph of our main page we have "It aims to describe all words of all languages using definitions and descriptions in English.". This is manifestly false. We do not include all words (we omit some brand names for example) and we do not treat computer programming languages to be languages. There are two main ways we could improve this situation. The first (my preferred option, as I'm sure you know) is to make the statement true - to include all words, in all languages. A second option is either to rewrite the statement as " ... most words of most languages ... " or to follow it with an asterisk that somehow points to a "terms and conditions apply" section. What does everyone else think? SemperBlotto (talk) 04:56, 7 June 2017 (UTC)Reply[reply]

What is this platonic definition of "word" that you seek to make us follow? Just because you say brand names are words doesn't mean we have to agree with you. DTLHS (talk) 04:59, 7 June 2017 (UTC)Reply[reply]
Of course you don't have to agree with me. But we define (deprecated template usage) brand name as a form of (deprecated template usage) name, and we define name as a type of (deprecated template usage) word. SemperBlotto (talk) 05:04, 7 June 2017 (UTC)Reply[reply]
Then you can see that it's totally impossible to "make the statement true" since we will never agree on what a word is. DTLHS (talk) 05:07, 7 June 2017 (UTC)Reply[reply]
This of course comes back to what constitutes a "word". If I stub my toe and say, "Yowzawhoawhoawhoa!" then that will certainly communicate some meaning to someone else ("that hurt a lot") but it's not really a word. There are a lot of signed, spoken, and written things which convey meaning but I think that anyone using common sense would realize that no dictionary could ever include all of these phenomena. —Justin (koavf)TCM 05:24, 7 June 2017 (UTC)Reply[reply]
"aspirational". --Catsidhe (verba, facta) 05:26, 7 June 2017 (UTC)Reply[reply]
Of course we would include "Yowzawhoawhoawhoa" - if it makes it into three different books by three different authors (like "Windows" has done). SemperBlotto (talk) 05:32, 7 June 2017 (UTC)Reply[reply]
Sure, and it never will. That doesn't make it more or less of a "word", it's just not a "word" for our purposes. Someone else could rightly call a lot of things "words" which we don't: everyone would have some caveats on what would constitute "every word in every language" and I don't think ours are so unreasonable as to expect them to need a disclaimer on the front page. —Justin (koavf)TCM 05:34, 7 June 2017 (UTC)Reply[reply]
For the record, maybe some dictionaries have a broader criteria for inclusion than us. http://jisho.org is an English-Japanese dictionary where in addition to "normal" stuff, you can search for some people names, brand names and movie titles, among other things. --Daniel Carrero (talk) 08:58, 7 June 2017 (UTC)Reply[reply]
  • Indeed, this is manifestly false: we only include attested words. Even if we relaxed attestation criteria even more, we still have to admit that we do not have an omnicorpus of all utterances of all languages that ever existed on the planet, and therefore, we will necessarily fail to cover some words. This is not just a hypothetical concern; we are very certain that we omit some words for lack of evidence, even though we do not necessarily have to know which words.

    One remedy is to do nothing and read the sentence as a slogan that does not contain the necessary qualifications. Another remedy is to inject "approximately", "basically" somewhere in the slogan. The asterisk mentioned above is also an option. Adding "as long as there is enough evidence", which occurred to me, is not so good since that would address the attestation requirement but not the other requirements and exclusions." --Dan Polansky (talk) 10:08, 7 June 2017 (UTC)Reply[reply]

  • We could change
    As an international dictionary, Wiktionary is intended to include “all words in all languages”.
    As an international dictionary, Wiktionary is intended to include basically “all words in all languages”, subject to certain conditions.
    --Dan Polansky (talk) 10:12, 7 June 2017 (UTC)Reply[reply]
    I like the proposed change. It's an honest assessment of what words we actually accept. Though technically we also accept phrases, symbols and other things. --Daniel Carrero (talk) 10:24, 7 June 2017 (UTC)Reply[reply]
  • How about "a bunch of words in a bunch of languages"? -WF
  • Footnoted motto/slogan? An example:
Our motto, annotated

All1 words2 in3 all4 languages5

The ordinary-word meaning of this slogan is somewhat misleading. The following notes explain the qualifications:

1Not every word is included at all, let alone in a meaningful way. Obviously we haven't gotten around to all of them. Attestation requirements exclude many. Due to the narrowness of our contributor base many languages are unrepresented and many specialized contexts are unrepresented, even in English.
2"Word" can include letters, numbers, symbols, abbreviations, proverbs, idiomatic expressions, some non-idiomatic expressions, clitics, affixes.
3Some "words2" could fall between languages. A multi-word expression borrowed from a foreign language could be non-idiomatic in its original language and thereby not includable in that language. It may also only be found in italics or quotation marks in running text in other languages, indicating that authors and editors don't think it has entered the lexicon in that language.
4See Vote on Serbo-Croatian.
5Translingual is not a language. Many non-words are better characterized as things. Things that are not words are not part of languages.

This approach works in real life for more-or-less unchangeable statements of great importance, like those in the US Constitution. DCDuring (talk) 13:09, 7 June 2017 (UTC)Reply[reply]

How about "Wiktionary: It's complicated." - [The]DaveRoss 13:31, 7 June 2017 (UTC)Reply[reply]
It seems to me that it worth expanding it at Wiktionary:All words in all languages, not as a disclaimer nor a policy but as an introduction to a pillar of Wiktionary. Comments on this thread worth to be annotated in a project/essay page, including "it's complicated" :-) --Vriullop (talk) 13:42, 7 June 2017 (UTC)Reply[reply]
I created Wiktionary:Votes/pl-2017-06/CFI leading sentence. --Dan Polansky (talk) 16:13, 16 June 2017 (UTC)Reply[reply]
I think the slogan is an acceptable aspirational slogan (good word, Catsidhe). Vriullop's idea of a page explaining it is interesting, but might overlap heavily with WT:CFI. It seems to be obvious to many people who comment on it that it is not to be taken to Amelia Bedelia levels of literalism. Even if we started including brand names, book titles, nonces, etc as some Wiktionaries do, we are unable to include all words in all languages, because some words were never recorded by anyone before they passed out of memory (e.g. in the Khazar language, Ciguayo, or Jassic). We are even apparently prevented by law from including all words in all languages because (in previous discussions in which some of our users who are lawyers have participated, it has been noted that) languages like Dothraki probably constitute significant parts of commercial franchises, and it would probably violate copyright if a third-party dictionary like us included all or a substantial number of the words in such a language. Even the most permissive inclusionism will hit hard limits.
Personally, I expect that as we become more complete, we will include codes for over 9000(!!!!1) languages. And we might broadly guesstimate that poorly-attested languages and highly-inflected languages may average out to half a million entries per language, so perhaps our slogan could say we aim "to describe four and a half billion words in nine thousand languages"? ;D - -sche (discuss) 17:44, 16 June 2017 (UTC)Reply[reply]

Sorry, I'm not sure now is the best time to create this vote since I had created another one two days ago. But we have an ongoing discussion about what to do with {{bor}} in all entries, so I created it anyway. Please check Wiktionary:Votes/2017-06/borrowing, borrowed. There are a few discussions linked there. Feel free to edit the vote or suggest any changes.

Aside from that, I intend to create a new Wikidata vote once the current one ends in June 11. --Daniel Carrero (talk) 14:10, 7 June 2017 (UTC)Reply[reply]

category for past forms of verbs used in turn as verbs on their own[edit]

A category of these forms may help the learner very much, since if they are not acquainted with these forms, finding them confuses momentarily the undertanding. For example, slew may mean "to veer", but it's as well the simple past tense of "slay"; likewise, "lay" is a transitive verb, as well as the simple past tense of "to lie, when pertaining to position". I do not know how to overlap categories so that I can get the one I wish to create. --Backinstadiums (talk) 14:12, 7 June 2017 (UTC)Reply[reply]

I think it would be useful to have a more general category for overlapping of verb forms. This category could also include set and read, which don't distinguish tense in writing. —CodeCat 14:17, 7 June 2017 (UTC)Reply[reply]
I see them as two different cases, the so-called "irregular verbs" being more treated than the one I propose. --Backinstadiums (talk) 14:47, 7 June 2017 (UTC)Reply[reply]
Could sb. please teach me how to proceed? --Backinstadiums (talk) 12:53, 8 June 2017 (UTC)Reply[reply]

WT:ELE - How to alphabetize languages[edit]

ELE dictates that language sections should be in alphabetical orders. Some languages have unusual characters in their English names, should they be alphabetized including those characters at face value, or without those characters? By way of an example, what is the correct order of "Ch'orti', Chachi, Cofán". - [The]DaveRoss 14:16, 7 June 2017 (UTC)Reply[reply]

I would say to just ignore the non-letters (from an English point of view). So the order would be Chachi, Ch'orti', Cofán. —CodeCat 14:17, 7 June 2017 (UTC)Reply[reply]
I find it hard to believe we haven't had this discussion before somewhere. I support using code-point order sorting, i.e. not ignoring the non-letters, since it's the easiest to implement and it's the state of all entries are in now. DTLHS (talk) 14:50, 7 June 2017 (UTC)Reply[reply]
If we decide to do something about this issue, let's please update WT:EL#Languages. It just says that languages besides Translingual and English are "in alphabetical order". --Daniel Carrero (talk) 14:52, 7 June 2017 (UTC)Reply[reply]
@DTLHS I figured the issue had been resolved and I was just not aware and couldn't find it readily. While I also like using the code-point ordering, I have found that not all entries are in that order (see: A).
@CodeCat If we go that way we will need to settle on what constitutes a non-letter (e.g. accents). Thankfully the set of allowable language names is limited so we can be comprehensive. - [The]DaveRoss 15:05, 7 June 2017 (UTC)Reply[reply]
One other note, the translation sections should also follow the same policy, whatever it is. And presumably other sorted lists should have a policy. - [The]DaveRoss 15:47, 7 June 2017 (UTC)Reply[reply]
I don't know about the examples you gave but I'd like to simply caution that those typographic characters actually are letters in some languages, e.g. ʻokina in Hawaiʻian. —Justin (koavf)TCM 16:03, 7 June 2017 (UTC)Reply[reply]
Right, but our L2 headers, i.e. language names, are in English. We have ==French==, ==German==, and ==Spanish==, not ==Français==, ==Deutsch==, and ==Español==. For that reason, I'm in favor of ignoring things like apostrophes and diacritics when it comes to alphabetizing languages. —Aɴɢʀ (talk) 20:39, 7 June 2017 (UTC)Reply[reply]
There is also a tendency to omit apostrophes in English when they don't seem to do anything (compare Mi'kmaq and Mikmaq), so the order might change depending on which spelling of a language we use, which is potentially confusing. Andrew Sheedy (talk) 20:49, 7 June 2017 (UTC)Reply[reply]
The list of L2 headers we currently employ includes many non-English characters, perhaps that should not be the case but it is at the moment. I think we are relatively consistent in this regard within a single language, but I might be wrong on that. - [The]DaveRoss 21:04, 7 June 2017 (UTC)Reply[reply]
I'm sympathetic to the argument that codepoint ordering is possibly the easiest to maintain (and is the one used by many entries now, due to bots sorting things), but it would not seem to be too difficult to define a more natural order, with Xârâcùù sorted before Xhosa, etc. It does appear as if other references ignore apostrophes, click letters, and diacritics when alphabetizing:
  • The International Encyclopedia of Linguistics lists in this order ’Akhoe, ǀAnda, Deti, ǁGana, Ganádi, ǀGwi, Hadza, Haiǀom, Hietshware, ǂHua (they print it as =/Hua), Juǀ’hoan [...] Nǀu, ǃOǃung, Sandawe, [...] ǀXam, ǁXegwi, Xiri, ǃXóõ.
  • Dalby's Dictionary of Languages sorts Larestani, Lārī, Lashi, Lāsī, Latgalian, [...] Mabwe-Lungu, Mača, Macao, [...] Māhārāshtri, Mahi.
  • The Ethnologue itself has Afrikaans, ||Ani, Birwa, [...] English, ||Gana, Gciriku, |Gwi, Hai||om, Herero, ‡Hua[sic], Ju|’hoansi, Kalanga [...] |Xam, ||Xegwi, ‡Ungkue.
  • Hodge's old Handbook of American Indians North of Mexico has Háami, Hāʼanaʟěnox, Haatse, Háatsü-háno, Habasopis, [...] Hailtsa, Haiʼ‘luntchi, Haiʼmāaxstō, Hai-ne-na-une, [...] Háiokalita, Haiowanni.
I agree that translations should be sorted in the same order.
Looking at WT:LOL, it appears that the list of characters besides A-Z used in language names and alt names (ignoring case) is
á, à, â, ä, ȁ, å, ã, ā : treat as a?
æ : treat as ae?
ɓ : treat as b?
ç, č : treat as c?
ḍ, ḏ : treat as d?
é, è, ê, ë : treat as e?
ɛ : also treat as e?
ğ : treat as g?
 : treat as h?
í, ì, î, ï, ĩ, ī, ɨ : treat as i?
ł : treat as l?
ñ : treat as n?
ŋ : treat as ng? At the moment, all languages with alt names using "ŋ" indeed use "ng" in their canonical names.
ó, ò, ô, ö, ȍ, õ, ō : treat as o?
ɔ, ɔ̃ : also treat as o?
š : treat as s?
 : treat as t?
ú, ù, ü, ũ, ŭ, ų : treat as u?
ŵ : treat as w?
ý : treat as y?
 : treat as z?
ə : what should happen to schwas?
() (as in "Kare (Africa)", "Yao (South America)") : parse as-is i.e. in codepoint order?
- (and , which was used in four alt names I just switched to use hyphens) : parse as-is?
. (as in "Mt. Iraya Agta") : parse as-is? or ignore i.e. discard?
', ʼ, ǀ, ǁ, ǃ, ǂ and ʻ and ˀ, ʔ (and the nonstandard ’, ‡) : ignore i.e. discard?
Note that many of these special characters only appear in alt names (which I included as a decent repository of what special characters might one day appear in canonical names of same languages), and are already normalized as above in their canonical names which we'd be dealing with, anyway! (Perhaps someone else feels like making a list of only those special chars which appear in canonical names.)
- -sche (discuss) 15:42, 8 June 2017 (UTC)Reply[reply]
Can we simplify your suggestion to: "Use the natural ordering after removing all combining characters and punctuation, and splitting ligatures."? I am worried about converting similar characters in other scripts to their Latin counterparts, since that way lies incredible complexity and subjectivity. - [The]DaveRoss 19:05, 9 June 2017 (UTC)Reply[reply]
My post only spells out all of the diacritical letters that are in use so people can see which letters those are and see if they agree with the proposed normalization; I expect that an actual rule would be phrased in more general terms, yes. For example, much of it can be simplified to "ignore diacritics". Noe links to an article by the person in charge of Glottolog about naming languages, which agrees with replacing (and sorting) ɛ and ɔ as e and o. As for punctuation, do we want to remove it? Suppose we had a language called "Kala (Zimbabwe)", should it be sorted before or after "Kala Lagaw Ya"? - -sche (discuss) 19:26, 9 June 2017 (UTC)Reply[reply]
I am certainly not the right person to make the calls about the best course here, I am merely hoping for a general rule if possible rather than a mapping system. - [The]DaveRoss 19:34, 9 June 2017 (UTC)Reply[reply]
I think parenthesized qualifiers such as "(Zimbabwe)" should be entirely ignored unless it results in two languages having identical names, and only then should they be used to sort them. In other words, sort it as just "Kala" but if there happens to be another "Kala" then use the qualifiers to determine which goes first. —CodeCat 19:36, 9 June 2017 (UTC)Reply[reply]
Wouldn't that be (effectively) just like sorting with the parenthesis left in place? Perhaps there are instances I am not considering. - [The]DaveRoss 19:48, 9 June 2017 (UTC)Reply[reply]
Perhaps, but editors can't be expected to know Unicode code points, whereas they can be expected to know English alphabetical ordering. So even if any programmed implementation treats it as you say, a human-readable description of the process would have to describe it more as I did. —CodeCat 19:57, 9 June 2017 (UTC)Reply[reply]
As far as I know, parentheses are only used when two languages do have the same name, and then, in almost all cases — the only exception that comes to mind is that we haven't yet added a qualifier to the million-strong language "Yao" just because the tiny, extinct language "Yao (South America)" exists — they are used on both languages. So I suppose we should leave the parentheses as-is and thus sort a hypothetical "Kala (Zimbabwe)" above "Kala Lagaw Ya". A rule that parentheses' contents "should be dropped unless X is true" where X is true 100% of the time would be needlessly confusing, IMO. - -sche (discuss) 20:00, 9 June 2017 (UTC)Reply[reply]
Not really, because "Kala (Zimbabwe)" should be sorted before a hypothetical language called "Kalaza". —CodeCat 20:09, 9 June 2017 (UTC)Reply[reply]
I think I may understand the source of the confusion. I'm assuming that a space doesn't count for sorting, since it's not an alphabetical character. So "Kala (Zimbabwe)" would be "Kalazimbabwe" for sorting purposes if we didn't take out the parenthetical part. Or, to take two real examples, I'm saying that "Tokelauan" would come before "Tok Pisin". —CodeCat 20:16, 9 June 2017 (UTC)Reply[reply]
Aha, that's a place our assumptions differed; I assumed spaces would be counted. Poking around other reference works, I see that G. Cinque's Typological Studies: Word Order and Relative Clauses sorts "Tokelauan, Tok Pisin" (in the alphabetical index), while J. Lynch's Pacific Languages: An Introduction sorts "Tok Pisin, Tokelauan". I'm not sure which is better. But independent of whether or not spaces are counted, I would never sort "Kala (Zimbabwe)" as "Kalazimbabwe". And if we both want "Kala (Zimbabwe)" above "Kalaza", isn't the simplest way to obtain that to treat the parentheses as parentheses, which are sorted ahead of alphabetic characters? - -sche (discuss) 20:49, 9 June 2017 (UTC)Reply[reply]
Here's a list of only those characters that are current used in canonical names, i.e. the ones we'd actually have to sort right now: [a-z], (space), - (hyphen), ' (apostrophe), . (dot), () (parentheses), á à â ä ã å ç é è ê ë í ì î ï ñ ó ò ô ö õ ú ù ü (diacritics, which could be handled by a rule "treat letters with diacritics the same as their base letters"), and ǀ ǁ ǃ ǂ (click consonants, which could be handled by a rule "treat click consonants as if they are not there"). It's possible that we should even remove some of those from the canonical names themselves, i.e. rename the languages. - -sche (discuss) 20:49, 9 June 2017 (UTC)Reply[reply]
Should hyphens be dropped? For example, how should Yan-nhangu, Yangkam, Yanomámi be sorted? - -sche (discuss) 22:52, 9 June 2017 (UTC)Reply[reply]
If we are doing this, I suggest that we create a module function that outputs a list of every language name in whatever internal order we decide on. Bots can read that page and order language sections and translations accordingly. DTLHS (talk) 22:57, 9 June 2017 (UTC)Reply[reply]
My suggestion- four classes of characters:
  1. Basic English letters
  2. Basic English letters with diacritics
  3. Non-English letters with no English counterpart (most or all them clicks and glottal stops)
  4. Punctuation
Perform the following transformations to produce the sort key, in the order given:
  1. Convert apostrophes to one of the other glottal-stop characters so they won't be treated as punctuation.
  2. Convert all punctuation to spaces and then convert multiple spaces to single spaces.
  3. Prefix all letters having diacritics with the corresponding basic English letter.
  4. Swap non-English/no-counterpart letters with the following letter so the following letter comes first. If there are multiple such letters, swap all of those letters next to each other as a group.
This has the advantage of having things sorted first by basic English letters, but having the order of the diacritics followed as well, and having the order of the "ignorable" non-English/no-counterpart letters followed, too. Since spaces are sorted before basic English letters, that also honors the principle that "nothing comes before something".
I'm not positive about the second transformation, since that will mean "Abc (def)" will be the same as "Abc def", but it's something to start with- tweaking is welcome. Chuck Entz (talk) 03:55, 10 June 2017 (UTC)Reply[reply]
@Chuck Entz, can you elaborate on transformation number four? - [The]DaveRoss 17:52, 16 June 2017 (UTC)Reply[reply]
Sure. The idea is that there should be a basic English letter before other characters for sorting purposes, with the others following it to distinguish between cases distinguishable only by those other characters. "Swapping" isn't really the best choice of words: what I mean is that the first basic English letter following one or more non-English-no-counterpart characters should be moved in front of them. Now that I've had a chance to think about it, maybe it would be better to ensure that the diacriticed letter goes with it, probably by moving the fourth transformation before the third, and clarifying that both basic English letters and diacriticed letters should be treated the same by this new third transformation. Thus 'ábçd would become aá'bcçd. Using these rules on -sche's examples below: Gadang→Gadang, Ga'dang→Gad'ang, and Madi→Madi, Ma'di→Mad'i. Chuck Entz (talk) 21:53, 16 June 2017 (UTC)Reply[reply]
...? Why is having to sort Mad'i any better than having to sort Ma'di? It seems like it would be neater to say: ignore apostrophes (clicks, and diacritics) when sorting languages on the page, but if that causes two or more languages to have the same name, then sort those two or more amongst themselves with the apostrophes present (left where they are).
Should spaces also be subjected to such a process (resulting in "Tokelauan, Tok Pisin"), or left alone? I tend to think spaces should be left in per the "nothing comes before something" principle you mention (so, "Tok Pisin, Tokelauan"), at which point there's no reason to remove the parentheses (and indeed, removing them would only make things more complicated and difficult), since leaving them in ensures that a hypothetical "Foo (Bar)" comes before "Foo Bar", which seems appropriate because bare "Foo" should also come before "Foo (Bar)". - -sche (discuss) 04:09, 17 June 2017 (UTC)Reply[reply]
You raise an important point, that some language names would be identical if diacritics and special characters were removed. These include gdk Gadang and gdg Ga'dang, and grg Madi and mhi Ma'di. What order should these be in: "Ga'dang, Gadang" or "Gadang, Ga'dang"? - -sche (discuss) 18:20, 16 June 2017 (UTC)Reply[reply]

Proposal: a page to centralise the patrolling effort[edit]

For some four years we’ve been unable to keep up with the rate of unpatrolled changes. This means that a lot of inadequate edits, and sometimes even vandalism, gets through. I’ve been trying to think of ways to improve the efficiency of our patrolling, because Special:RecentChanges is very hard to use: it lists all unpatrolled edits in all languages and all areas, but an individual patroller has the knowledge to patrol perhaps 10% of them. For example, right now most recent unpatrolled edits are changes to Hungarian entries and the addition of Galician, Ukrainian and Italian translations. I have enough knowledge to verify whether the Galician translations are good, and I could look up some published dictionaries to check the Italian and Ukrainian translations (but other users could do it faster and better), and I can’t possibly hope to check the content of the Hungarian entries (just the formatting).

In addition to the raw recent changes page, we could have a page with a list of users with unpatrolled edits, separated by language and topic. For example, if I am patrolling the recent changes and come across a user adding Japanese etymologies, I would add a new item to section ==Japanese==, subsection ===Etymology=== on this page, with a link to the user’s contribution page and perhaps an explanation as to why I think their contributions need special attention. Eventually a patroller who is more proficient in Japanese etymology will see this link and check the contributions. In order to keep the patrolling process invisible, as it already is, this page should have a mechanism to prevent users from being pinged (WT:VIP has such a mechanism, if I remember correctly). The advantages that such a page might bring include:

  • Encourage users who don’t the time or patience to go through Special:RecentChanges to patrol.
  • Encourage patrollers to delegate edits to someone who feels more confident in the language and topic.
  • Provide a place where the correctness of someone’s edits can be discussed (like a pre-WT:RFC, without implying that their edits need to be cleaned up)
  • Make it easier to identify sockpuppets and patterns of odd behaviour.
  • Prevent unpatrolled edits from being lost in the Recent Changes limbo.

Ungoliant (falai) 16:38, 7 June 2017 (UTC)Reply[reply]

This is a fantastic idea. I think that the problem of a piling up of unpatrolled edits has not been due to the difficulty of patrolling so much as to the fact that not that many people are patrolling. This kind of page could help make clear to admins who don't patrol as much how they can help out in the common effort. —Μετάknowledgediscuss/deeds 18:39, 7 June 2017 (UTC)Reply[reply]
I think that this collation of edits by topic and language could be done mostly automatically if anyone wants to do it. DTLHS (talk) 18:41, 7 June 2017 (UTC)Reply[reply]
I am not sure I am understanding the suggestion exactly, but it does seem like it could be useful. Is there any way you could mock something up which demonstrates what you are suggestion (even if in a limited way)? I think I would support this effort even if that wasn't possible, I am more curious about what the implementation might look like and be capable of. - [The]DaveRoss 13:27, 8 June 2017 (UTC)Reply[reply]
@TheDaveRoss mockup. — Ungoliant (falai) 15:22, 8 June 2017 (UTC)Reply[reply]
Thank you for that, it isn't exactly what I expected but it makes a lot of sense now. This seems like a great way to collaborate. - [The]DaveRoss 15:38, 8 June 2017 (UTC)Reply[reply]
What did you have in mind, Dave? — Ungoliant (falai) 15:59, 8 June 2017 (UTC)Reply[reply]
For some reason I pictured something which showed the edits to be patrolled, and I couldn't think of how that might work (without a lot of fancy coding). I get now that it is more of a WT:VIP analog. - [The]DaveRoss 17:21, 8 June 2017 (UTC)Reply[reply]
@Ungoliant MMDCCLXIV: I patrol changes to existing Hungarian entries almost every day by checking my Watchlist. Since it doesn't show new entries, those are harder to find. It would be great to have a better method. We had a Recent Changes list by languages a long time ago, that worked well. I'm not sure what technical challenges it presents. --Panda10 (talk) 18:45, 8 June 2017 (UTC)Reply[reply]

Appendix: Easily confused chinese words[edit]

Hi, regading the concept of chinese anagram, I think it would be of great help to create an appendix similar to Easily_confused_Chinese_characters but Easily "confused Chinese words", which in theory could be easily created from a corpus of words, just selecting those with the same number of the same characters yet in different positions. Furthermore, I'd like to know how the concept of anagram can be used for characters themselves, transposing radicals or even strokes. --Backinstadiums (talk) 13:14, 8 June 2017 (UTC)Reply[reply]

Wikidata precautionary principle[edit]

Once the Wikidata vote ends, there's a chance we'll get Wikidata installed here.

If that happens, what do you think of restricting its use by implementing the rule below?

"Any and all edits using Wikidata shall be reverted on sight if they were not discussed or voted before."

Or we might demand all Wikidata uses to be voted, not just discussed. --Daniel Carrero (talk) 14:31, 8 June 2017 (UTC)Reply[reply]

I agree that it is right to be cautious in any implementation of data from Wikidata, especially when that data will be presented directly to the user. I would suggest that, at least to begin with, any use of Wikidata data which is non-controversial (e.g. using the mapping between ISO codes and language names [in as much as they conform to our current use]) be discussed publicly and agreed upon, and any potentially controversial use (e.g. including a Wikidata identifier [Q12345] as a template parameter; presenting Wikidata data directly) be subject to a vote. - [The]DaveRoss 14:45, 8 June 2017 (UTC)Reply[reply]
I'm not sure the language code→name mapping is 100% noncontroversial. I've been thinking it might be a good idea to create a separate vote with this proposal: "Moving all the data from Category:Language data modules, Category:Dialectal data modules, Module:families/data, Module:scripts/data and Category:Unicode data modules to Wikidata." --Daniel Carrero (talk) 15:01, 8 June 2017 (UTC)Reply[reply]
Ehh. (Or maybe: LOL.) Treatment of languages (as dialects or macrolanguages, etc) is rather complex, and not a good candidate for quick Wikidata-fication. Treatment of ISO- as well as exceptionally- coded lects as separate or as dialects of one unit varies a lot between wikis; we merge even some lects that have separate wikis, like the Serbo-Croatian lects. Names also vary a lot not just between wikis of different languages (which obviously use their own native names for things), but also within different wikis that use the same language, if wikis have different priorities with respect to e.g. calling each language by its native name, or by the name that is most common in references on the language, vs calling it by a name that distinguishes it from other languages with the same name: hence we have "Austronesian Mor" and "Mbo (Congo)", where another wiki might prefer "Mor (Austronesian)" or "Mbo (Democratic Republic of the Congo)", or might even think the name should be plain "Mor" and damn the torpedoes.
Other things that pertain to languages also vary between wikis, e.g. some wikis might consider that a language that is documented by linguists in the Latin script, or in Cyrillic, but that has no natively-used script should not be said to have "Latn"/"Cyrl" as a script, whereas other wikis might feel otherwise. Since our templates need to know what scripts a language is written in so as to know whether it needs a transliteration, and since our CSS applies fonts based on that information as well, we wouldn't want other folks to pull the rug out from under us as to what script a language had.
Even family information is the source of both disagreement (is Tibeto-Burman different from Sino-Tibetan? is Finno-Ugric different from Uralic? etc) and different priorities (some wikis might want comprehensive family trees that listed every node; for us, they would be too finely granular and would ghettoize languages and derivations into tiny categories hidden at the ends of long trees).
To move only some language names / script infos / etc to Wikidata and handle others locally would be inefficient and unwise, IMO. And in order to move them all, we would need "infrastructure" to be added there that would handle e.g. "is called X on wiki Y", at which point, we'd just be doing the same thing we're doing well here, but doing it there for some reason, which seems inefficient and unwise.
IMO, treatment of languages is so central to what each Wiktionary does that, especially for a wiki with as many active linguistically-adept and technically-adept editors as en.Wikt, it makes sense to do it locally.
Peripheral things, things that are not core to our mission, are better candidates for moves: e.g., parsing that a certain city is in a certain county/province/etc in a certain country on a certain continent, or (re the recent discussion of eponyms) parsing that a certain person has a certain nationality. - -sche (discuss) 16:34, 8 June 2017 (UTC)Reply[reply]
Here's an idea: if Wikidata allows it, each language could have an "English Wiktionary name" property. We would use "Austronesian Mor" even if for other purposes the same language is called by other names. --Daniel Carrero (talk) 16:54, 8 June 2017 (UTC)Reply[reply]
There is a strong argument for bringing ourselves more into compliance with international standards where we can, but also I disagree that managing by exception would overly complex. There are relatively few people who have any idea how/where to modify the existing language mapping structure, it is both obfuscated and has significant functionality issues (see water, man). As Daniel suggests there are also other paths for this (and virtually every) problem, e.g. creating new properties which would be controlled by this community. - [The]DaveRoss 17:01, 8 June 2017 (UTC)Reply[reply]
I'd like to say something again about compliance with standards, which is important. I'm pretty sure all or most language codes listed at Module:languages/datax are not in compliance with ISO. I would support changing them all to be ISO-compliant, but some people may oppose doing that. (this was one of the points discussed at Wiktionary:Beer parlour/2017/February#Proposal: Implementing Wikidata access) --Daniel Carrero (talk) 17:10, 8 June 2017 (UTC)Reply[reply]
  • The sentence in quotes is something that I would enthusiastically support. It would certainly assuage a lot of my (and others') personal fears with respect to Wikidata. —Μετάknowledgediscuss/deeds 15:19, 8 June 2017 (UTC)Reply[reply]
I support requiring a vote for each distinguishable use case. DTLHS (talk) 16:47, 8 June 2017 (UTC)Reply[reply]

Wiktionary:Votes/2017-05/Installing Wikidata passed. I created Wiktionary:Votes/pl-2017-06/Wikidata precautionary principle to implement the principle proposed here. In the vote, I rewrote the proposed text to be a bit longer and more policy-like in my opinion, but the idea is 100% the same as proposed here. Feel free to edit the vote or suggest any changes. --Daniel Carrero (talk) 00:30, 11 June 2017 (UTC)Reply[reply]

1000 Middle Dutch entries![edit]

With the appropriately-chosen fêeste, there are now 1000 entries for Middle Dutch. Some of them are alternative forms, but this is compensated by some entries that have more than one lemma. —CodeCat 20:52, 8 June 2017 (UTC)Reply[reply]

Impressive! Our coverage of languages like this is an asset. - -sche (discuss) 19:16, 9 June 2017 (UTC)Reply[reply]

Borrowed descendants[edit]

Is there is standard format for words whose descendants are all borrowed? Case in point, French hangar. Related, I wonder if we need a variant of {{see desc}} for examples like Frankish *haimgard that instead reads (see there for further borrowings). --Victar (talk) 00:53, 9 June 2017 (UTC)Reply[reply]

What do you care for most? What are you concerned with? Take part in the strategy discussion[edit]


The more involved we are, the more ideas or wishes concerning the future of Wikipedia we have. We want to change some things, but other things we prefer not to be changed at all, and we can explain why for each of those things. At some point, we don’t think only about the recent changes or personal lists of to-dos, but also about, for example, groups of users, the software, institutional partners, money!, etc. When we discuss with other Wikimedians, we want them to have at least similar priorities that we have. Otherwise, we feel we wasted our time and efforts.

We need to find something that could be predictable, clear and certain to everybody. A uniting idea that would be more nearby and close to the every day’s reality than the Vision (every human can freely share in the sum of all knowledge).

But people contribute to Wikimedia in so many ways. The thing that should unite us should also fit various needs of editors and affiliates from many countries. What’s more, we can’t ignore other groups of people who care about or depend on us, like regular donors or “power readers” (people who read our content a lot and often).

That’s why we’re running the movement strategy discussions. Between 2019 and 2034, the main idea that results from these discussions, considered by Wikimedians as the most important one, will influence big and small decisions, e.g. in grant programs, or software development. For example: are we more educational, or more IT-like?

We want to take into account everybody’s voice. Really: each community is important. We don’t want you to be or even feel excluded.

Please, if you are interested in the Wikimedia strategy, follow these steps:

  • Have a look at this page. There are drafts of 5 potential candidates for the strategic priority. You can comment on the talk pages.
  • The last day for the discussion is June, 12. Later, we’ll read all your comments, and shortly after that, there’ll be another round of discussions (see the timeline). I will give you more details before that happens.
  • If you have any questions, ask me. If you ask me here, mention me please.

Friendly disclaimer: this message wasn't written by a bot, a bureaucrat or a person who doesn't care about your project. I’m a Polish Wikipedian, and I hope my words are straightforward enough. SGrabarczuk (WMF) (talk) 11:01, 9 June 2017 (UTC)Reply[reply]

Yikes, the use of marketing jargon here is horrible, the opposite of wikis' origins as common-sense technical tools. I think I agree with the Germans. Equinox 18:50, 9 June 2017 (UTC)Reply[reply]
@Equinox: Excellent thanks for the link! --WikiTiki89 19:03, 9 June 2017 (UTC)Reply[reply]
I wish they would keep us out of their propagandistic marketing agendas. --Victar (talk) 19:08, 9 June 2017 (UTC)Reply[reply]
This isn't marketing. I'm not a marketer. And I'm not 'we'. Tell me you don't care about anything beyond English Wiktionary, fine, I'll understand (there are many users who 'just edit'), but don't imply I do things that I don't. For me, it's an utter lack of WT:AGF. In other words: are you interested in the movement strategy? do you have any questions? remember: questions to me personally, not to the entire WMF. And please, see my user page and read who I am. SGrabarczuk (WMF) (talk) 21:16, 10 June 2017 (UTC)Reply[reply]
I apologize, SGrabarczuk (WMF), the principle of WT:AGF has never been very popular on en.wiktionary. Don't take it personally, though. Plenty of our first-time visitors get jumped on. Think of it as baptism by fire. —Stephen (Talk) 22:01, 10 June 2017 (UTC)Reply[reply]
Stephen, I appreciate what you wrote, however, I'm not new at all. I got used to a daily MMA situation when my opponents get medieval on my arguments, and I can do the same with their words. That's a 'normal' wiki-way, but only when one doesn't like the opponent, or when one talks to someone from the other side (a newbie, a WMF staffer). But I'm not one of them, and there's no reason not to like me in advance. I, like 'Red' Redding, may get your community sth from the other side, provided I'm asked civilly. That propaganda, I wrote it myself. I tried to avoid corporate speech, added a friendly disclaimer, but yeah. Cheers. SGrabarczuk (WMF) (talk) 00:52, 11 June 2017 (UTC)Reply[reply]
SGrabarczuk, one of our admins has indefinitely blocked w:Jimmy Wales himself. We are nothing if not ecumenical abusers. —Stephen (Talk) 20:58, 16 June 2017 (UTC)Reply[reply]
No, nobody is attacking you for no reason. Stuff like "movement strategy ecosystems and actors" is opaque marketing-style jargon from the commercial world, not to be trusted on an open project. Equinox 21:05, 16 June 2017 (UTC)Reply[reply]
@Equinox, SGrabarczuk (WMF) is Polish, and his English can be expected to be a bit off. It likely reflects the sort of English texts that he often reads. He indicates that his English level is en-3, so judgments such as "opaque marketing-style jargon from the commercial world, not to be trusted on an open project" are unfair, inappropriate, and very likely incorrect. —Stephen (Talk) 23:08, 16 June 2017 (UTC)Reply[reply]
@Equinox, I bet both of us well know how to deal with the History tab. You can see that I didn't write that. You'll welcome to react like tell (err, where did you find this?) insert_the_author's_nick_here not to use that sort of words. I'll agree! I know that most of users assume that guy has a (WMF) in his signature, so he's co-responsible for all that rubbish, but that's an incorrect, simplifying, and not exactly fair assumption. I'm not a boss, thus I'm responsible only for myself. SGrabarczuk (WMF) (talk) 22:05, 16 June 2017 (UTC)Reply[reply]
@Stephen G. Brown Really, I always found en.Wikt to be very manageable and relaxed in terms of good faith. Make a round on German Wikipedia (and German editors here...). Quick to assume agendas and deceit and not as shy to say so. (Not excluding myself, I quietly accuse a lot of them of agenda pushing.) @SGrabarczuk (WMF) I agree that the headlines sound like suit buzzwords. The actual explanations are valid topics that might benefit the Wiki projects. I think editors, certainly here, have little interest in diverting time and attention to the peripherals of what they actually like doing. En.Wikt is, from what I see, populated not by a people who wanted to be part of a commune and better the world but by people who use dictionaries a lot, do linguistics a lot or have a pet language they would like to propagate. Small wik-, big -tionary. Even the participation in these here talk pages is minuscule. This means for the 'other side' that things have to be made with the awareness/anticipation that they will be judged (and most likely discarded) after a cursory glance, so things like headlines should be very plain and on point, even if it doesn't have much ring as a motto. And to give you at least some form of feedback: The only things I ever see in these discussions which people would need other than 'more editors making more edits' are technical things. So maybe explicitly asking what's needed in that regard would get a better reply. Really just a guess, though, I don't deal in the Grease Pit. Korn [kʰũːɘ̃n] (talk) 22:45, 16 June 2017 (UTC)Reply[reply]
@Korn, read the comments in this discussion by Equinox. Not what I would call manageable and relaxed in terms of good faith. —Stephen (Talk) 23:08, 16 June 2017 (UTC)Reply[reply]
@Stephen G. Brown Might be personal bias as I'm partially affected, but I still find "Marketing jargon isn't to be trusted in an open project" less venomous than the things I find on the talk pages of Wikipedia, and I don't think that kind of attitude is common here towards editors editing. I could only tell you one person here who ever showed an actual lack of good faith [read: was instantaneously making blatant accusations] in dealing with such situations and it's a German. Even Wyang and CodeCat only think of each other as stubborn idiots, rather than people with ill intent. As an example of actual bad faith: My worst personal encounter with not-good-faith was moving an article called 'German dialects', dealing in equal parity and no little detal with Dutch, Low German and German dialects, to 'Continental West-Germanic dialects'. I was, IIRC, then accused of doing that for vandalism, by people who had never heard the term, and the fact that I had made a typo, as I do, and had to move the article twice, was taken as evidence that I was covering up my villainy. The article was then moved back and butchered from a perfectly good article about Continental West-Germanic into a disgrace about 'German'. Now that is real lack of good faith, the important kind, which affects the projects' contents. And I never saw that kind happening here. (Feel free to direct me to examples.) Equinox' distrust, while not a good sign of project spirit, doesn't actually make Wiktionary any worse to the user than it is, nor does it drive away able, good-willed editors. And I don't think he's actually thinking WMF is out to worsen the projects. I read it more as an accusation of being out of touch. Korn [kʰũːɘ̃n] (talk) 09:00, 17 June 2017 (UTC)Reply[reply]
@Korn, yes, not everyone treats others that way. The number of abusive editors seems to have been decreasing over the years, very gradually. Nowadays, I think there are only a couple of them left. In the old days, half of our editors seemed to be looking for any reason to jump down someone's throat. The atmosphere was tense. It used to be common for some admins to block other admins over minor or weird reasons. A few years ago, a certain admin blocked an new editor for one year because he transcribed a Persian word using /sh/ instead of /š/. The situation is improving, but it is still nothing to brag about. —Stephen (Talk) 09:22, 17 June 2017 (UTC)Reply[reply]
It's no insult. All the propaganda and stuff coming from "the Wikimedia foundation" uses a completely different tone and context than the ENTIRE rest of the Wiktionary site. I see it as inappropriate. Contributors here may be involved in the Wikipedia project or other WM projects as well, but are not necessarily and many aren't. Wikipedia and Wiktionary are two completely separate communities, other than the fact that the two often link to each other on a lot of pages, often share contributors, and that they're funded by the same foundation. In my opinion, besides possible funding issues, if Wiktionary ever broke off from its relations with WMF, there would be no differences. The people, the contributors, the tone, expectations of contributors, style of editing, etc., would all be the same. It is my opinion that the ads need to stop coming to places like the Beer Parlour because it's simply not a concern or priority of ours. PseudoSkull (talk) 10:43, 17 June 2017 (UTC)Reply[reply]
Really, now? You're attacking the people who do the friendly work of making sure this project stays online because they offer you the choice to participate in the decisions of what they do with the funds they raised? Which you can just ignore and move on with your life? That's the type of thing you want to do? Korn [kʰũːɘ̃n] (talk) 12:28, 18 June 2017 (UTC)Reply[reply]
As above, I'm not the only one holding the side you just stated. Also, I wouldn't say my statement is "attacking", but rather stating an opinion that the propaganda is somewhat inappropriate here. Maybe we should just make a page called Wiktionary:Place propaganda and advertisements here to keep it out of our site discussion space. Like I said, Wiktionary is a completely separate community with very little direct relations to Wikipedia, so the propaganda is not really a concern to us. That isn't an attack; that's just the stating that I'm annoyed by it like others above me, as it has always clearly stood out to me that the propaganda did not belong here. I never said anything, though, because I didn't want to possibly be that one egg who said something that pissed Wiktionary off. PseudoSkull (talk) 16:54, 18 June 2017 (UTC)Reply[reply]
To be fair, I never said that the content of the propaganda was bad; I personally actually agree with most of what it said and think it's good that free knowledge is being highly encouraged by the foundation, etc., but the fact that there IS propaganda at all is the bad part, regardless of how much I agree or disagree with its statements. PseudoSkull (talk) 16:57, 18 June 2017 (UTC)Reply[reply]

Maybe there's a story in the background that justifies your attitude, I don't know. I probably can't and shouldn't resolve that. I can only admit that I understand what you write. Now, look: in various places all over the world, groups of people (not necessarily users) gathered and talked about the future of our movement. Link #1, link #2. The crucial points documented in such pages will help to conduct next the discussion. Think and write what you want, but don't miss the point that you do have an opportunity and you can influence the strategy, if you take part.

By the way, the next cycle (3rd one) of our discussions will begin in July. Now, we're reading the cycle 2 feedback. When the conclusions are published, I'll write again. Do you have any questions concerning the strategy? SGrabarczuk (WMF) (talk) 15:30, 24 June 2017 (UTC)Reply[reply]

To clarify: I really, really dislike marketing/advertising and feel that they have destroyed much of what was good about the early Internet (and I spend enough time deleting Viagra and vanity from en.wikt); so I am inclined to go on the attack when I think I spy this stuff creeping into spaces that are still largely neutral and informational, like (the greater part of) the Wikimedia projects. SGrabarczuk, of course I have nothing against you personally and I apologise if you feel you were being attacked as an individual. I know I am difficult to get along with! Equinox 15:45, 24 June 2017 (UTC)Reply[reply]

English orthographic categories[edit]

Hi, taking into account the educational characteristic inherent in lexicography, I'd like to propose creating categories for intricate issues related to orthography. Thus, a category for words with doubled letters, even twice (lemmas: aggress, accommodate, etc.) would help learners review lists and make less mistakes. --Backinstadiums (talk) 16:34, 9 June 2017 (UTC)Reply[reply]

We have a few already: Category:English terms by orthographic property. — Ungoliant (falai) 16:48, 9 June 2017 (UTC)Reply[reply]

Wikidata proposal: Add "English Wiktionary name" as a statement about all languages[edit]

@Lea Lacroix (WMDE): Here in the English Wiktionary, we use a single name for each language everywhere.


The chosen language name is used in etymologies, translations, descendants, categories, appendices, policies and other places. If we decide to change a language name, it will have to change everywhere. We often use modules and templates to convert language code→name information, such as "mhz"→"Austronesian Mor". We have large data modules containing language code→name information. The "mhz"→"Austronesian Mor" information is one of the 621 languages currently stored at Module:languages/data3/m. See also Category:Language data modules for all the language data. WT:LANG is our language policy.

Three questions:

  1. Can we move all the language data modules to Wikidata? (we don't know yet if there's consensus here to do it — it may or may not be done, depending on consensus)
  2. Can we add a new statement "English Wiktionary name" where its value is exactly the name we use? This way, the language name will be exactly as the English Wiktionary decides, even if a different synonym is chosen for other projects and purposes.
  3. Can we protect the new statement "English Wiktionary name" so that people are generally disallowed to edit it even if the rest of the data item is free to be edited? Maybe only Wikidata admins would edit it.

This concern was raised by @-sche in this discussion above: #Wikidata precautionary principle. --Daniel Carrero (talk) 17:23, 9 June 2017 (UTC)Reply[reply]

The canonical name of a language is the single most queried item across Wiktionary. This means that there are very strict requirements on speed. How fast can such names be queried from Wikidata? What if it's hundreds of them like on water? As for the name of the data item, something like en.wiktionary might be better, but I have very little knowledge of how Wikidata works in detail (could someone explain it to me on my talk page?). —CodeCat 17:35, 9 June 2017 (UTC)Reply[reply]
@Daniel Carrero, Lea Lacroix (WMDE): If this foolish idea is to proceed, then as I noted in the previous discussion, you would also need to add "English Wiktionary script(s)" and "English Wiktionary family" field, and corresponding "French Wiktionary script(s)", "English Wikipedia family", etc to handle cases where script and family information is disputed/controversial between wikis, e.g. where one wiki follows Ethnologue in declaring a language unwritten (Zxxx / Zyyy, because there is no natively-authored literature) but we regard it as being in Cyrl (if we have entries in Cyrl citing linguistic reference works in Cyrl, our templates and CSS rely on that being declared as a script of that language so they know to ask for transliteration, and to call on the right fonts for displaying it); and (respectively) where one wiki treats Uralic and Finno-Ugric as different (maybe a WP would), and another treats them as the same (cf Tibeto-Burman vs Sino-Tibetan, etc), or where one wiki wants every node in a family tree represented, while another wants pragmatic levels of categorization, not to ghettoize languages into a dozen nested levels of families. You would also need parameters for when one wiki considers several ISO or non-ISO coded lects to be distinct, but another wiki considers them one language (e.g., Serbian vs Croatian vs Serbo-Croatian, Rhine Franconian vs Central Franconian vs Kölsch, European French vs Cajun French, Boubonnais-Berrichon vs Bourbonnais and separately Berrichon etc), as well as the basic capacity to handle lects which lack ISO codes. You would need, in effect, to duplicate everything we currently do here, but on another site, where the people who maintain language data on this site would need to have the full editing rights to keep it up to date there, or bother Wikidata editors about hundreds of things a month — see how many distinct codes I've changed in Module:languages's submodules recently — or, more likely, have things go unupdated. And, as CodeCat notes, fetching the data from Wikidata would have to be as fast as fetching it from our local modules, unless you were going to make the existing problems large entries have (which seem to be due to the use of many complex auto-transliteration modules) worse. I continue to regard this fixation on outsourcing everything, despite the lack of benefits, as foolish; language data is so central to each wiki that it makes sense to do it locally, at least on a wiki with as many active technically- and linguistically-adept editors as en.Wikt. - -sche (discuss) 18:10, 9 June 2017 (UTC)Reply[reply]
I'm curious as to whether using Wikidata would actually be faster than our local language modules. Is there any way to measure that? Currently, a template replacing "mhz"→"Austronesian Mor" would have to transclude the entirety of Module:languages/data3/m. With Wikidata, it would only have to access a single property (language name, or en.wikt language name) of d:Q2122792. --Daniel Carrero (talk) 20:03, 9 June 2017 (UTC)Reply[reply]
Arbitrary access to Wikidata is an expensive function, that is access to a page not connected with the current one. See mw:Extension:Wikibase Client/Lua#mw.wikibase.getEntity. I think it is not an option for the water case. --Vriullop (talk) 20:51, 9 June 2017 (UTC)Reply[reply]
That makes me wonder: can we have per-wiki Wikidatas? We don't want to outsource all this data, but the infrastructure of Wikidata could still be beneficial (whether in terms of speed remains to be seen) if it could be kept locally. —CodeCat 20:19, 9 June 2017 (UTC)Reply[reply]
It's important that d:Q42365 (the Wikidata item for "Old English") already has statements identifiers like "Quora topic ID" ("Old-English") and "Encyclopædia Britannica Online ID" ("topic/Old-English-language"), so using Wikidata to contain information about multiple sites seems the normal thing to do, it's not an earth-shattering new idea. So I think Wikidata might as well have "English Wiktionary name" (or "English Wiktionary ID") for all languages, it would simply fit the current system. --Daniel Carrero (talk) 20:26, 9 June 2017 (UTC)Reply[reply]
Perhaps, but what about my idea of having some kind of Wikidata local to en.wiktionary alone? —CodeCat 20:41, 9 June 2017 (UTC)Reply[reply]
OK, about your idea: Support. --Daniel Carrero (talk) 20:42, 9 June 2017 (UTC)Reply[reply]
Sounds to me that what we need is actually WikiMetadata: a database of controlled vocabularies (and even ontologies) used by one Wiktionary to store and serve all linguistic descriptions, be it languages or parts of speech or scripts. No actual "data", only edited by recognized Wiktionarians, cached for fast Lua access, etc. — Dakdada 11:22, 14 June 2017 (UTC)Reply[reply]
More and more I think the idea of a local Wikibase installation might be the better course. While I like the idea of sharing as much data as possible across Wiktionaries, I am afraid that the consensus to get there is not likely to arise any time soon. If we start with a local database we can get all of the benefits of a relational database to work with while eliminating the concerns of numerous individuals. - [The]DaveRoss 17:55, 16 June 2017 (UTC)Reply[reply]


Thanks for bringing this interesting issue. I'll try to provide a general answer, feel free to ping me if I forgot something.

I'm not sure that a "English Wiktionary name" property would be accepted through the community process on Wikidata. Of course I can't answer on behalf of the editors. In order to get the languages names from Wikidata, maybe it would be more efficient to first improve Wikidata (we are aware that a lot of items about languages are missing), fix the labels if necessary (in collaboration with the Wikidata editors), and finally improve your module so you can use data from Wikidata, but also, in the cases where Wikidata doesn't fit to your naming rules, use your own local labels.

About having part of the data protected: I'm pretty sure this will be rejected by the community, since Wikidata, as the other Wikimedia projects, has the free and direct editing in its basic rules. We don't allow any editor, or external organization, to have specific rights on the data, and we try to solve the potential issues with discussions, source-based informations. I'm sure we can solve this as well whithin both communities.

About having a local database: it is technically possible that you install your own instance of Wikibase, the free software that powers Wikidata. However, our goal here is to share knowledge whithin the Wikimedia projects as much as possible, and try to have less informations split into silos, that's why we can't support this idea.

Now I understand better your concerns about languages, I have the feeling that this is not the best topic to start experimenting with Wikidata and the arbitrary access. These templates are used on almost all the pages of the main namespace. It would be wise to start with something with a smaller scale of change. Also, this topic appears quite controversial sometimes. I agree that we should find solutions together, but for a start, I would suggest something else.

I had a look at the citations namespace, for example this one, and noticed that you're generally including with the quote, the name of the work, its author, and date of publication. These informations could be very easily integrated from Wikidata. Instead of entering manually "1843 — Charles Dickens. A Christmas Carol", you could build a small module that needs only the ID of the work (Q62879) to display automatically the title, author, year of publication, and even more informations. This seems a nice way to start experimenting with arbitrary access. What do you think? Lea Lacroix (WMDE) (talk) 13:57, 12 June 2017 (UTC)Reply[reply]

@Lea Lacroix (WMDE): Thanks for your reply. I understand. I agree with the citation idea, I've been thinking that it's a great idea to use Wikidata to fetch that information (and of course help to build Wikidata by adding data about more books when needed). --Daniel Carrero (talk) 02:05, 14 June 2017 (UTC)Reply[reply]
About the local database: the idea is that this would not be a database of shareable data, but of metadata. This is very different. Also those metadata are already "split into silos" (in Lua modules), and for good reasons: they are community specific. — Dakdada 11:27, 14 June 2017 (UTC)Reply[reply]

Request to add a new language code "rya" for rGyalrong a.k.a. Jiarong[edit]

I'm travelling in rGyalrong speaking areas of Sichuan right now. Danba county and now Kangding, both in Garze (Ganzi).

rGyalrong people identify as Tibetan and are classified as Tibetan by the Chinese government. But there language is more closely related to the Qiang language than to Tibetan. (The Qiang don't idenity as Tibetan and are classified separately by the Chinese government.)

There is an ISO code 'jya'

rGyalrong is apparently very important in reconstructions of Old Chinese as it's considered to be a very conservative member of the Sino-Tibetan family.

Note that it's considered a group of languages (or dialects?) but a proposal to split it into individual language codes was rejected in 2011. This can be handled with labels and in any case I believe none have written forms so we'd have to use either IPA or whatever conventional orthography is used by linguists. One main linguist is known for the study of all these languages so my guess is there's a unified conventional orthography.

hippietrail (talk) 04:14, 10 June 2017 (UTC)Reply[reply]

-sche split the code "jya" last year into Situ (sometimes called Eastern rGyalrong) (sit-sit?), Japhug (sit-jap?), Tshobdun (Caodeng, Sidaba) (sit-tsh), Zbu (Rdzong'bur, Showu, Sidaba) (sit-zbu). DTLHS (talk) 04:24, 10 June 2017 (UTC)Reply[reply]
If you can obtain more data on how different or similar the lects are, perhaps especially when written, that will be very helpful. The limited data I found on the lects suggested they were not mutually intelligible; even the Ethnologue page says "Dialects are likely three separate mutually unintelligible languages" with low similarity. Guiillaume Jacques says "Rgyalrong comprises at least four mutually unintelligible languages: Japhug, Tshobdun, Zbu, and Situ." That's why I proposed the split DTLHS links to, and (with no feedback for over a month) implemented it. - -sche (discuss) 05:58, 10 June 2017 (UTC)Reply[reply]
Thanks for the feedback! None of them are regularly written though I read that Situ had an orthography created before 1950 I think. I haven't found any info on that yet so I guess it's something made up by linguists and/or missionaries. Situ is also by far the most spoken. It's also the one spoken in both Danba and Kangding, so it's the one I'm interested it.
Here is the best technical info I've found so far. I'm still reading through it: http://www.academia.edu/969613/Rgyalronghippietrail (talk) 11:45, 10 June 2017 (UTC)Reply[reply]
I've been gathering some comparisons of the languages' pronouns, conjugation patterns and other words at User:-sche/Gyalrong. The last decade or two of literature seems to be in agreement that the lects are mutually unintelligible. The only user I can think of who hasn't commented but might know about these languages is @Wyang. - -sche (discuss) 22:59, 11 June 2017 (UTC)Reply[reply]
@-sche I'm not an expert in Rgyalrong either, and only have a physical copy of the 2002 Chinese-Rgyalrong dictionary (in the Situ dialect). The western expert on Rgyalrong is definitely Dr. Guillaume Jacques, who also used to have an account and was an admin on the Chinese Wikipedia: w:zh:User:向柏霖, so it may be a wise idea to contact him re: the organisation of the varieties of Rgyalrong on Wiktionary. Dr. Jacques wrote the 472-page 《嘉绒语研究》 (Jiarongyu yanjiu, “A study on the Rgyalrong language”), which divided Rgyalrong into the four mutually unintelligible dialects above. I also vaguely remember there was a Rgyalrong-Chinese-French dictionary for the Japhug dialect circulating online, which may be handy. Wyang (talk) 08:16, 12 June 2017 (UTC)Reply[reply]

Flag semaphore[edit]

I created three flag semaphore entries. Let me know if they look OK and if they should be kept. I also asked SemperBlotto (User talk:SemperBlotto#Flag semaphore).

I tried to imitate the notation used in Category:American Sign Language lemmas.

(It may be of interest that Morse code entries were created in 2016, they also fit the spectrum of "things you can use in place of letters and numbers". The Morse discussion is here: Wiktionary:Beer parlour/2016/August#Proposal: Creating entries for Morse code characters.) --Daniel Carrero (talk) 12:34, 10 June 2017 (UTC)Reply[reply]

@Daniel Carrero: Thanks for this. With Braille, Morse Code, and semaphore, I think that covers most unusual encodings for the Latin alphabet other than fingerspelling and shorthand. —Justin (koavf)TCM 16:56, 10 June 2017 (UTC)Reply[reply]
@Daniel Carrero, Koavf: Tactile Sign Language: at least its alphabet should be added to Wiktionary. I don't know whether it should be a a type of fingerspelling or rather a new language as such http://www.deafblindinformation.org.au/wp-content/uploads/2016/01/db-tactile-alphabet.pdf
Thanks in advance. --Backinstadiums (talk) 08:31, 11 June 2017 (UTC)Reply[reply]
@Backinstadiums: I knew that tactical signing was a thing but not that there was a way to encode it in print. It seems like this is a pictorial chart and not a way to record it that we can use. And tactile signing is not a language itself but another encoding. —Justin (koavf)TCM 15:31, 11 June 2017 (UTC)Reply[reply]

Should this be a redirect?--2001:DA8:201:3512:CD84:BF8E:5FA5:70A7 16:54, 11 June 2017 (UTC)Reply[reply]

Yes, looks good to me. (single codepoint) and II (two instances of "I") are the same thing, even though there's a distinction at some level which is meaningful to computers. For the same reason, I had redirected say, to !. --Daniel Carrero (talk) 17:00, 11 June 2017 (UTC)Reply[reply]
Ok, I won't revert anymore. It's a kind of instinct when you see an IP user doing wholesale deletion of stuff, you think something's fishy. —CodeCat 17:02, 11 June 2017 (UTC)Reply[reply]
IP person, if it's not too much trouble, please add {{R character variation}} in these kinds of redirects! --Daniel Carrero (talk) 17:06, 11 June 2017 (UTC)Reply[reply]
Could you write documentation and categorise the template, @Daniel Carrero? —CodeCat 17:07, 11 June 2017 (UTC)Reply[reply]
Alright, done! --Daniel Carrero (talk) 17:19, 11 June 2017 (UTC)Reply[reply]

How can we use Wikidata's existing data pool?[edit]

It's very controversial to adopt Wikidata for anything new that's specific to Wiktionary, but is there any data currently on Wikidata that we can already make use of? Data about species comes to mind. @DCDuring, what do you think? —CodeCat 17:03, 11 June 2017 (UTC)Reply[reply]

Author / biographical information is what comes to mind for me. DTLHS (talk) 17:04, 11 June 2017 (UTC)Reply[reply]
Certain "is-a" categories, e.g. Rome is a city, labrador is a dog. Equinox 17:06, 11 June 2017 (UTC)Reply[reply]
  • Support for types of place names. --Daniel Carrero (talk) 17:07, 11 June 2017 (UTC)Reply[reply]
    • The place name thing is interesting. We might be able to modify {{place}} to make use of it. However, we'd presumably need some way to tell, within an entry, "use this Wikidata item". Would that mean that {{place}} would take a parameter to specify the Wikidata item code (Q...)? —CodeCat 17:10, 11 June 2017 (UTC)Reply[reply]
      • I think we should think about if we can avoid using numeric codes directly in entries, if that's possible. DTLHS (talk) 17:13, 11 June 2017 (UTC)Reply[reply]
        • If that's possible, then sure. But what is the alternative? —CodeCat 17:14, 11 June 2017 (UTC)Reply[reply]
          • I suggest using a combination of numerical codes and hidden text comments that don't affect the entry. d:Q90 (Paris, France) d:Q79917 (Paris, Arkansas) could work like in the table below. --Daniel Carrero (talk) 17:29, 11 June 2017 (UTC)Reply[reply]
Code Result
# {{Wikidata place|Q90|capital of France... this is just a comment I can say anything here}}
# {{Wikidata place|Q79917|city of Arkansas... this is the same}}
  1. A city in Île-de-France, France and the capital and most populous city of France.
  2. A city in Arkansas, USA.
Bleh. If we're going to include comments, can we not put them in actual wikitext comments instead of a pointless template parameter? —CodeCat 17:37, 11 June 2017 (UTC)Reply[reply]
I think I would be happy enough without any comments at all, just using the 1st parameter for the number code and that's it.
But if we want comments for all place names, I was hoping to do this: if the comment parameter is empty, the entry could be categorized in Category:Place names without comments. --Daniel Carrero (talk) 17:40, 11 June 2017 (UTC)Reply[reply]
I like the idea of using comments since it gives an assurance that the wikidata item is actually the intended target- otherwise someone could add an incorrect number and there would be no way to tell what they meant or if it was wrong. DTLHS (talk) 17:45, 11 June 2017 (UTC)Reply[reply]
The data itself might help. The module could check (somehow) if the item is in fact a city, town, river or some other kind of geographical thing. If it's not, then it could throw an error. —CodeCat 17:48, 11 June 2017 (UTC)Reply[reply]
Two checks: 1) check if the item is a type of geographical location (presumably needed for description and categorization purposes, and a template called {{Wikidata place}} should be able to return a module error otherwise); 2) compare the current entry title with the accepted titles in the Wikidata item. When "Q90" and "Q79917" are used in the entry Paris, the module should be able to check if "Paris" is an acceptable name for both items as per Wikidata. --Daniel Carrero (talk) 17:57, 11 June 2017 (UTC)Reply[reply]
That second check might not work. The template is used for definitions in all languages, so it might also be used on Dutch Parijs. —CodeCat 18:01, 11 June 2017 (UTC)Reply[reply]
Dutch Parijs is already available in the list of names for d:Q90 in all languages. This reminds me, {{Wikidata place}} should be able to know what is the current language section, so in that Dutch entry the proposed syntax should actually be {{Wikidata place|nl|Q90|capital of France}} (and "en" in English entries, of course). --Daniel Carrero (talk) 18:06, 11 June 2017 (UTC)Reply[reply]
That works, but what if the name isn't listed? Should we require the name in language X to be present in Wikidata before we allow the use of the template for X? Adding a tracking category ("hey, this name isn't in Wikidata yet, someone go add it!") would be much more helpful than a straight error. —CodeCat 18:10, 11 June 2017 (UTC)Reply[reply]
I support adding a tracking category as you described. Maybe it could be called Category:Dutch place names missing in Wikidata or something. --Daniel Carrero (talk) 18:14, 11 June 2017 (UTC)Reply[reply]
A top-level category for stuff to be added to Wikidata in a particular language would also be good. Since you were involved in renaming all the request categories, I'll leave the naming to you. :) —CodeCat 18:17, 11 June 2017 (UTC)Reply[reply]
Alright! Proposed category tree (which may be changed/discussed):
If we have template tracking categories as discussed here, I believe we don't actually need those template comments. Sure, it would be wrong to set up a place name definition with the number d:Q10943 because it means "cheese", but the template should be able to recognize that automatically. --Daniel Carrero (talk) 18:36, 11 June 2017 (UTC)Reply[reply]
Hmm, wouldn't they rather be request categories? —CodeCat 19:15, 11 June 2017 (UTC)Reply[reply]
I and other people in the request category vote seemed to support the following notion: a request category is when you manually request something, like {{rfe}}. In the Wikidata categories above, the entries would get automatically categorized whenever something looks wrong. --Daniel Carrero (talk) 19:18, 11 June 2017 (UTC)Reply[reply]
Ok. But to call it an error is a bit extreme. It's just a missing translation, a common symptom of a project that is always a work in progress. —CodeCat 19:20, 11 June 2017 (UTC)Reply[reply]
OK, second proposal:
--Daniel Carrero (talk) 19:36, 11 June 2017 (UTC)Reply[reply]
I don't like this since it gets into the entire problem of representing lexicographic data in wikidata which they are really not set up to do. Place names have synonyms, obsolete forms, dialectal forms, etc. DTLHS (talk) 18:18, 11 June 2017 (UTC)Reply[reply]
This is purely to supplement the current {{place}} template. This template creates definitions based on the parameters you give it and then also categorises appropriately. What would change in a Wikidata implementation is that these parameters would be fetched from Wikidata (e.g. "is a city", "capital of France") rather than being provided as parameters. This isn't lexicographical data in my understanding of the word. —CodeCat 18:22, 11 June 2017 (UTC)Reply[reply]
Knowing that Dutch Parijs is a synonym of English Paris is lexicographic data. DTLHS (talk) 18:24, 11 June 2017 (UTC)Reply[reply]
And that follows from the fact that both uses of the {{Wikidata place}} template, one on Paris and one on Parijs, would use the same Wikidata item code. This information is therefore not stored on Wikidata at all. Even if there were no Wikidata and {{Wikidata place}} were an empty template, the mere fact that they both had Q90 as a parameter would establish them as synonyms. —CodeCat 18:29, 11 June 2017 (UTC)Reply[reply]
It's important that the "city of France" sense in Dutch Parijs will need to access the English translation somehow. Here's two ways to accomplish that: using Wikidata or using a parameter. See table. --Daniel Carrero (talk) 19:04, 11 June 2017 (UTC)Reply[reply]
Code Result
# {{Wikidata place|nl|Q90}} <!-- using Wikidata -->
# {{Wikidata place|nl|Q90|Paris}} <!-- using a template parameter -->
  1. Paris (city in Île-de-France, France and the capital and most populous city of France)
The current iteration of {{place}} uses t1= for that purpose. I think we should keep using a parameter, again to minimise our dependence on Wikidata for lexicographical things. —CodeCat 19:07, 11 June 2017 (UTC)Reply[reply]
That is an option, but it seems Wikidata is still an option too because it will get "morpheme" data items designed specifically for lexicographical data (Wikidata:Wiktionary).
Maybe the best course of action is just keep using the parameter as you said since it's reliable and it works. But we can change our minds later and delete that parameter from all entries if the morpheme thing works out. --Daniel Carrero (talk) 19:11, 11 June 2017 (UTC)Reply[reply]
Yes, I'm aware of how hesitant many people here are about offloading lexicographical data to Wikidata. But if we limit ourselves to the existing data already out there, such as topography in the case of {{place}}, then I don't think it would be as much of an issue. —CodeCat 19:14, 11 June 2017 (UTC)Reply[reply]
In my role as articulate fish at a convention of ichthyologists, here's my initial view of how Wikidata might help with Translingual taxonomic entries.

How about a dynamic map for place names? See ca:Kenya. It fact it does not use currently Wikidata but OpenStreetMap with Wikidata identifier. With Wikidata access it could fetch coordinates to add a point in the OSM map, for example for cities. --Vriullop (talk) 12:20, 12 June 2017 (UTC)Reply[reply]

The map could be included in an infobox with relevant links to Wiktionary. From d:Q114: capital=Nairobi, demonym=Kenyan, languages=Swahili, English, currency=shilling, TLD=.ke, ISO code=KE. --Vriullop (talk) 13:12, 12 June 2017 (UTC)Reply[reply]
Wikidata might be a desirable way for me to speed insertion of "References" to external sites. I have seen such links on Commons, en.WP, and Wikispecies. Wikidata seems to have already accumulated such information from multiple projects, though I haven't yet found which ones. If that were better than what I could find on eg, Rosa at NCBI, then it would marginally speed things up. If I could extract the links in a single step by, say, a substed template (or something more sophisticated), that would save a more meaningful amount of time. I am skeptical about the response time of accessing such links through Wikidata each time an entry is loaded.
Something similar might apply with respect to references at vernacular names, though such references are, IMO, not so useful for such entries.
You might think that the hierarchical taxon/clade structure would be a perfect use of Wikidata, but I believe that relying on such a structure is not helpful in definitions. It seems better to me to define species, genera, tribes, and sections with reference to the family, not matter what intermediate ranked taxa or clades may be used by one or more sites. Non-expert users don't seem to have much familiarity with the various super-, sub-, infra- ranks of phylas, classes, orders, families, and species, let alone tribes, sections, and divisions. DCDuring (talk) 19:49, 11 June 2017 (UTC)Reply[reply]
See for example w:ca:Poecilia latipinna. At footpage a template fetches the identifiers from d:Q906572 and links to databases. --Vriullop (talk) 12:12, 12 June 2017 (UTC)Reply[reply]
That's the idea, but [] . See [[Alconeura]] for a not uncommon approach to handling missing pages at external databases: providing a link to a page for a higher-ranked taxon. I suppose that could be managed by using a separate template for each higher-ranked taxon selected for inclusion, but I'd want to be able to exclude from the higher-taxon list of links those databases that had a more specific link. In principle there could be many such separate templates for a given taxon, though very few taxa would need more than three. DCDuring (talk) 14:21, 12 June 2017 (UTC)Reply[reply]
I'm not sure if I understand. d:Q10404513 has 3 identifiers linking to 3 databases. Are these links ok? Anyway you must provide always the Wikidata page to fetch. Either this one or the higher-ranked. --Vriullop (talk) 14:44, 12 June 2017 (UTC)Reply[reply]

Wikidata items as senseids[edit]

Currently, the template {{senseid}} is used to disambiguate specific senses and allow us to link to them from elsewhere. Senseids are just text, they can be anything at all as long as it's unique. This means it's possible to use the codes for Wikidata items as senseids too. For example, we could use {{senseid|en|Q90}} on the first sense of Paris. This wouldn't actually do anything, other than tell editors that this sense refers to the thing whose Wikidata code is Q90. {{senseid}} would not be modified at all for this purpose, it would not access Wikidata. But it does add information to Wiktionary entries, by establishing a conceptual link between senses and Wikidata items.

Such links could be used for future things that we currently haven't thought of. A possibility that comes to mind is Wikipedia links. If {{senseid}} detects that its parameter is a Wikidata item code (Q followed by numbers), then it could be modified in the future to query that item for its en.wikipedia article name. {{wikipedia}} or some similar interproject link could then automatically be displayed next to certain senses, whenever {{senseid}} is given a Wikidata item as a parameter.

For those of us hesitant to offload data onto Wikidata, please notice that this change would change nothing at all on Wikidata's end. The data added by this would be entirely on en.wiktionary, in the form of a wikitext template call, so we have full control. No data would be added to Wikidata, we'd merely be using what's already there. —CodeCat 18:47, 11 June 2017 (UTC)Reply[reply]

I've added Wikidata senseids to Paris (English only) to demonstrate what I mean. Wikidata isn't actually enabled yet on Wiktionary, so these do literally nothing other than provide a senseid anchor on the page. —CodeCat 11:54, 12 June 2017 (UTC)Reply[reply]
Looks good to me. Two comments: 1) doing this in large scale would probably require a vote, 2) I guess this idea should work 100% well for place names, but we can't use Wikidata items as senseids for all kinds of Wiktionary definitions. Wikidata probably doesn't have separate items for each sense of the verbs do, have, go, be... I guess the future Wikidata "morpheme" thing should work, but apparently the database would need to be built from scratch. --Daniel Carrero (talk) 12:03, 12 June 2017 (UTC)Reply[reply]
Indeed, such senseids would probably be limited to nouns only, as Wikidata doesn't currently have items about verbal actions. The important thing to keep in mind is that the Wikidata items are about the referents of words, not the words themselves. The words would have their own items, if we get around to that. This does create some issues that I foresee. The colour green has Wikidata item d:Q3133. But in our entry green, we have not only a noun referring to this colour, but also an adjective. Since senseids must be unique within a single language section, we can't give both of them {{senseid|en|Q3133}}. So which one do we put it on? —CodeCat 12:10, 12 June 2017 (UTC)Reply[reply]
Using Wikidata senseids would probably work well, not only for place names but also for a lot of proper nouns. For common nouns, adjectives, verbs and everything else, I don't have actual numbers, but at first sight I think it would fail more often than not. d:Q9465 is "ethics", so would we use it which sense of ethics and/or ethical, if any? d:Q7242 is "beauty", and our entry beauty has multiple related senses too. --Daniel Carrero (talk) 12:23, 12 June 2017 (UTC)Reply[reply]
We could just add disambiguators to the ids. They are still just strings, after all. So {{senseid|en|Q3133-noun}} for the noun green would work too. As long as the Wikidata id can still be parsed out, it should be fine. As for ethical, I think the issue here is that ethics is a field of study, which the adjective doesn't really have much to do with. If it did indeed refer to the same thing as ethics, I see no reason not to include the Wikidata id there too. —CodeCat 12:52, 12 June 2017 (UTC)Reply[reply]
I'm fine with adding disambiguators to the ids, with hope that later in the future we can replace all that stuff by the Wikidata "morpheme" thing. --Daniel Carrero (talk) 12:56, 12 June 2017 (UTC)Reply[reply]
I've added a tracking template, Special:WhatLinksHere/Template:tracking/senseid/Wikidata, to {{senseid}} whenever a Wikidata id is used as a senseid. This allows us to keep track of them for the time being. The process of adding all these senseids to entries will be a long one. I've thought of a possible way to speed it up, though, once Wikidata is enabled. Module:headword can be modified so that it checks if a Wikidata label exists with the same name as the current page. If so, it could track the page. This would give us a list of all pages that could probably have a Wikidata senseid added to them. —CodeCat 19:01, 13 June 2017 (UTC)Reply[reply]
Nice work! We should use senseid / senselinks more often, but as you said they are quite tedious to add at the moment. A gadget could also be an option, with a Wikidata suggest-style dropdown. – Jberkel (talk) 16:52, 14 June 2017 (UTC)Reply[reply]

Note that senses will have a new ID in Wikidata lexeme data model: mw:Extension:WikibaseLexeme/Data Model, with format L3746552-S4. --Vriullop (talk) 10:59, 17 June 2017 (UTC)Reply[reply]

This is one of the uses that I feared, having wikitext littered with numerical identifiers. Ugly. --Dan Polansky (talk) 15:21, 17 June 2017 (UTC)Reply[reply]

Deverbative or deverbal?[edit]

User:Barytonesis recently created a template {{deverbative}}. This is a good idea but I think it's wrongly named and should be {{deverbal}}. Both terms are synonyms but "deverbal" is much more common (as well as shorter and easier to type) — about 9x as many hits in Google, plus my spelling checker actually marks "deverbative" (but not "deverbal") as a mistake, plus Wikipedia has entries for Deverbal noun and Deverbal adjective but no entries for deverbative anything, not even redirects. The template puts these terms under the category "Foo deverbatives" (none of which have been created yet, and should not be created at all probably). I think instead they should go under "Foo deverbal nouns" or "Foo deverbal adjectives"; this requires an optional pos= parameter (which should default to "noun"). There are under 50 entries currently using this template and all appear to be nouns. I can easily use a bot to rename the template uses. Any objections? If not I will go ahead and make the changes. Benwing2 (talk) 23:37, 11 June 2017 (UTC)Reply[reply]

I agree. It's also consistent with denominal (for which there should also be a {{denominal}} template). Additionally, it should point to Appendix:Glossary. --Victar (talk) 01:07, 12 June 2017 (UTC)Reply[reply]
Fine by me. I'm actually happy that you take interest in it --Barytonesis (talk) 09:59, 12 June 2017 (UTC)Reply[reply]
While, you're running a bot, you should change the outdated |lang= to |1=. --Victar (talk) 10:13, 12 June 2017 (UTC)Reply[reply]

I hereby nominate User:Dan Polansky for administrator on the English Wiktionary, with the one condition that Dan will be disallowed from using the block tool. For some context, quoting Dan himself:

"For context, my admin vote failed in August 2016.

Note that the block tool is one of power over other people, and should not be awarded to people who we cannot trust, no matter how good editors they are. The use of the block tool does not require consensus, and blocks are rarely challenged. Multiple of current admins are not qualified to use the block tool, in my view. The deletion tool can be abused to hide trails of conversation; it was used in this way by an English Wiktionary admin who meanwhile ceased editing."

I believe Dan would be able to delete pages in accordance to the rules and such, and seems to have a need for such. Still not sure about the block tool, but I'm throwing this condition in because it puts some previous opposers at ease a bit. Honestly, a lot of users may most of all fear that Dan may abuse the block tool. I think Dan is a great editor, though he may be rude sometimes and is notorious for such, so the admin tools of page deletion at the very least should be awarded to him. IMO, the admin tools are not supposed to be awarded to people because they're "nice" (and I'm not necessarily saying Dan is not nice), but because it would be useful for them to have to make even more constructive contributions, for instance, by fighting obvious vandalism, deleting pages in accordance to RFV and RFD, etc.

As I've never started a vote before, who wants to start this vote? PseudoSkull (talk) 18:22, 12 June 2017 (UTC)Reply[reply]

I am somewhat uncomfortable with the idea that someone be given user rights that they are then prohibited from using. Either they should have them with the trust of the community to use them appropriately, or not have them. If there are, as Dan suggests, multiple users who should not have the blocking tools but should be able to delete (and protect?) pages, we could always create another user group with just the rights which are applicable. We could, for instance, give folks in the "template editor" group delete rights, or create a "deleters" group with just the rights to delete and restore. It is even possible to have delete rights without the ability to see deleted revisions, etc. I don't know if the complication is worth the extra effort, but to me it is preferable to the alternative. - [The]DaveRoss 19:04, 12 June 2017 (UTC)Reply[reply]
I am also uncomfortable with that idea. I would probably support creating a new group "Deleters". --Daniel Carrero (talk) 02:08, 14 June 2017 (UTC)Reply[reply]
Along with another group for "non-deleters". --Victar (talk) 02:33, 14 June 2017 (UTC)Reply[reply]
Non-deleters would be anyone not in the "deleter" or "administrator" groups. - [The]DaveRoss 11:07, 14 June 2017 (UTC)Reply[reply]

Local groups[edit]


Yesterday, I went to met my local colleagues, contributors of Wikipedia, Wiktionary, Commons, Wikidata, OpenStreetMaps or OpenFoodFacts. We met several times this year, to discuss about contribution and drink beers. Is there some contributors here that do the same in their local groups? If yes, do you also contribute on Wikipedia and chat about this project or do you give news from Wiktionary to Wikipedians. It's not a sociological inquiry, it's just curiosity. I have no idea how local groups works out of France. Noé 10:06, 13 June 2017 (UTC)Reply[reply]

I think that the concentration of francophone Wikimedians in a relatively small, easily accessible area (France) is quite different from the anglophone community. I just recently met another Wiktionarian and would be happy to meet more, but it isn't likely to happen often. I also have plans to meet some Wikipedians later this year and maybe speak a bit about Wiktionary, although I honestly don't know what I'd tell them. —Μετάknowledgediscuss/deeds 17:34, 13 June 2017 (UTC)Reply[reply]

Unprotect WT:NFE[edit]

I think it's silly that this is protected against non-admins. Surely non-admins also have important news to announce? —CodeCat 19:27, 13 June 2017 (UTC)Reply[reply]