Template talk:IPA/Archive 13

Remove fonts for XP?
Template Unicode will be deleted due to being obsolete. As XP is no longer supported, its font declararion (targeting XP) will also be removed. That begs the question: is the current font declaration for IPA (also targeting XP) still necessary, or can they be removed as well? Note this would only impact XP users, and the template itself is not under discussion; only the font. 18:57, 11 April 2016 (UTC)
 * No response... I asume that means no objections. Since it affects XP only, which we no longer support, I will remove the font declaration.  18:02, 25 April 2016 (UTC)

error in French example
§ Conversion to IPA says




 * French, IPAc-fr ||  →
 * }

That's wrong. French $⟨j⟩$ represents the voiced postalveolar fricative [&zwnj;ʒ&zwnj;], but IPA $⟨j⟩$ represents the voiced palatal approximant.

And for that matter, where's the conversion chart for the "natural spelling"? It sure isn't at Template:IPAc-fr/doc (where I'm also raising this question) or Help:IPA for French. How do you get /ʒ/ in IPAc-fr?

Cross-referencing from WikiProject Linguistics.

Please me to discuss. --Thnidu (talk) 18:13, 25 April 2016 (UTC)

Module:IPA
I've created this module. It behaves the same way as this template when only one parameter is used, but if a language code is added to the first parameter, it creates a link to that language's help page – for instance,  becomes /ɛɡ.zɑ̃pl/. The list of help pages in the module isn't complete, but once it is, I think this module could be used to give this template a bit more functionality. — Eru·tuon 02:11, 17 October 2016 (UTC)

Finished the list. All ISO 639-1 and ISO 639-3 language codes for the languages that have IPA keys are included. — Eru·tuon 04:24, 17 October 2016 (UTC)

Render g as ɡ
Hi Wikipedians, Would it be possible to render a /g/ as a /ɡ/ (the correct one) in the template for convenience?? Anyone who listens is the best!!😃 Thanks. —Awesomemeeos (talk) 06:37, 7 July 2016 (UTC)
 * At the moment, that would be somewhat clunky to do replacements (replace) would have to be used), but the module I'm working on (Module:IPA) could easily do that (as well as converting from X-SAMPA to IPA). — Eru·tuon 01:51, 17 October 2016 (UTC)


 * IPA/sandbox now replaces the regular g with the IPA (script) g, using Module:IPA:
 * &rarr; [g]
 * &rarr; [ɡ]
 * Unfortunately, it can't change the text that's supplied to the template; it just makes sure the correct g is displayed. — Eru·tuon 23:52, 17 October 2016 (UTC)

Kazi Zafar Ahmed
Kazi Zafar Ahmed is displaying "Lua error in Module:IPAc-en at line 13: attempt to index local 's' (a nil value)". That appears to be a result of recent edits to this template, although possibly the error is old wikitext in the article. Would someone please work out what is going on. Johnuniq (talk) 06:09, 25 July 2017 (UTC)
 * The IPA template was used to link to an audio but didn't include a notation, which obviously isn't the intended use of the template. Nardog (talk) 09:28, 25 July 2017 (UTC)
 * Module:IPAc-en should be modified to prevent this happening in future. Lines 180 and 182 should contain  to check if the parameter in question exists before trying to use string functions on it. — Eru·tuon 16:14, 25 July 2017 (UTC)

How can we call this template from the "IPA-xx" templates without sending brackets?
Statement of the problem

Not even sure that this is correct place to ask this.... My immediate concern is IPA-fr and IPA-mg, but this a situation endemic to all "IPA-xx" templates where "xx" is some language code. Since IPA-fr etc. call this template (IPA) this seems like a good centralized place to discuss this.

IPA-fr sends a string ("tananaʁiv" for example) to (IPA. Well not exactly. If I specify the the string, "tananaʁiv", it actually sends "[tananaʁiv]". Note the brackets. But I don't want the brackets.

We could edit IPA-fr and IPA-mg to solve my immediate problem (I can't do it as I lack the chops to do it correctly). But if we did that it would leave IPA-fr and IPA-mg being different from all the other "IPA-xx" templates, which is not good. We'd have to change all the "IPA-xx" templates.

I could, in IPA-fr and IPA-mg, add to the switch statement (that is, it's treated like a blank parameter), then in the call use "nobrackets" (if present) to send "string" rather than "[string]" to the IPA. This seems kludgy.

I guess better would be to send a parameter to here and then strip the brackets here... I guess maybe even easier would be never send them, and then add the brackets here in the absence of a param telling it not to...? Above my pay grade.

Or I may be missing a simple solution that's right there. If so, then Bob's your uncle.

But if not, what is our best way forward here?

Specific instance showing why this is a problem

The article Antananarivo begins thus:

This is objectively horrible on several levels. So what we want to do is change the lede sentence to something that actual humans can read:

and then later in the article body be like

but as you can see the brackets are carried into the text by IPA-fr and IPA-mg, and we want to strip them. I mean, they're not horrible there, but they don't belong either, and it's non-excellent to have them.

(I do not know why "tananaʁiv" and "antananaˈrivʷ" are presented to the reader instead of the more atomized individual sounds I usually see. I assume that it's because these are non-English pronunciations and that is proper and necessary and why we have IPA-fr etc. rather than just using IPA.)

So now what? Herostratus (talk) 20:12, 6 October 2017 (UTC)
 * Why in the world would we want to srip them? The brackets indicate what is enclosed is a phonetic notation as opposed to standard text. See the article International Phonetic Alphabet. Nardog (talk) 20:32, 6 October 2017 (UTC)
 * What he said. There's also a huge difference between transcriptions enclosed within brackets and transcriptions that are enclosed within slashes. We could safely switch to slashes in Dutch, Serbo-Croatian, Polish, Hungarian, etc. by the way, but that's another issue. Mr KEBAB (talk) 00:40, 7 October 2017 (UTC)

Template-protected edit request on 19 June 2017
The IPA-tagging function in Module:IPA is ready to be deployed widely. It creates the exact same content as the existing template, but also changes regular g to IPA ɡ. Please replace the content of the template with the following: — Eru·tuon 22:01, 19 June 2017 (UTC)
 * You might want to see about implementing this one. --Izno (talk) 18:52, 21 June 2017 (UTC)
 * Are there test cases showing the difference in behavior does not exist? --Izno (talk) 18:53, 21 June 2017 (UTC)
 * I created testcases at Module talk:IPA/testcases. It compares the output of to the output of, which currently uses  . The first example differs only because it changes the   to  . — Eru·tuon 20:55, 21 June 2017 (UTC)
 * FYI, module test cases are usually designed to test the Lua functions while template test cases show the changes in the final output of the templates. You might consider jiggering those pages. If no-one comes along to review the nill difference in a day or three, turn answered back to n. --Izno (talk) 21:39, 21 June 2017 (UTC)
 * I don't know what you mean by "template test cases". I'm not aware of any way to compare the results besides using Lua. I suppose I could replace the template with the actual Lua function in the module, to prevent it changing if someone decides to use the sandbox template for something else. — Eru·tuon
 * Oh, I suppose you mean this. — Eru·tuon 22:20, 21 June 2017 (UTC)

See the template test cases at Template:IPA/testcases. Output of the sandbox template is identical to the output of, except that the non-IPA symbol   is changed to. — Eru·tuon 21:28, 22 June 2017 (UTC)
 * ✅. Nardog (talk) 21:41, 22 June 2017 (UTC)

I have reverted the change because an error was reported at IPA-hu. It looks like the "g" in the link to Help:IPA for Hungarian was transformed to "ɡ" (the IPA symbol) as a result of the change. We should fix it. Nardog (talk) 22:40, 22 June 2017 (UTC)
 * Oops. I hadn't thought about links. I'm working on the module now. — Eru·tuon 22:58, 22 June 2017 (UTC)
 * The module now finds links and only does replacements in the displayed part of the link. — Eru·tuon 23:21, 22 June 2017 (UTC)
 * Now it also ignores the contents of HTML tags. It should be reliable now. — Eru·tuon 23:39, 22 June 2017 (UTC)
 * Could you change the template back to using the module again? — Eru·tuon 16:52, 23 June 2017 (UTC)
 * The input could include any kind of string that includes "g" that shouldn't be replaced, such as other templates and unpiped links. You still could escape those by adding more codes, but now I'm thinking... Is this worth the effort? Isn't this something, say, a bot should be taking care of? As much as I do think we need a meta template to cover IPA templates for all languages, the substitution of "g" might just not be within the scope of such a template. Nardog (talk) 17:21, 23 June 2017 (UTC)
 * Well, I could add a tracking category for un-IPA  that is not inside HTML tags or templates or targets of piped links, instead of replacing it. Should I do that? — Eru·tuon 17:33, 23 June 2017 (UTC)
 * Sure, that is a much safer approach. I haven't looked deep into the code, but I think what we should do is first preproccess the input and then avoid links and HTML tags. That way you can distinguish 's inside &#123;&#123;...&#125;&#125; that are supposed to appear in the outcome and ones that are not. Sorry if I'm preaching to the choir. Nardog (talk) 17:45, 23 June 2017 (UTC)
 * Preprocessing shouldn't be necessary, and in fact it could mess things up, as template parameters are already preprocessed. (I did a test at wikt:Module:User:Erutuon/sandbox.) — Eru·tuon 17:56, 23 June 2017 (UTC)
 * I've implemented tracking. Not sure if the tracking category is in keeping with whatever template-tracking categories already exist. — Eru·tuon 18:07, 23 June 2017 (UTC)
 * adds the category, but  doesn't. I don't think it's ready. Nardog (talk) 18:17, 23 June 2017 (UTC)
 * Oops. That was due to an unnecessary if-statement. Fixed. — Eru·tuon 18:22, 23 June 2017 (UTC)

Does using the right kind of "g" really justify a whole new lua module? Oppose as unnecessary over-complication &#123;&#123;3x&#124;p&#125;&#125;ery (talk) 19:28, 21 June 2018 (UTC)

A bug
If the template ends up being at the end of a line of rendered text and the stress falls on the first syllable of the word, the line break may be inserted between the and. For example:


 * / ˈbˠaːiːxt̪ˠ/

Instead of:


 * /ˈbˠaːiːxt̪ˠ/

Can someone please fix this? Esszet (talk) 20:32, 9 August 2018 (UTC)
 * That's not a bug so much as an intended behavior of your browser. A slash triggers a line break after it, and a stress triggers a break before it. You can avoid it by writing . Note this would also prevent breaking at spaces. If you still want the notation to break at a space, write e.g.   or  . Inserting  after , or before   or  , as in  , is yet another option.
 * If we were to convert IPA to a Lua module, we should definitely add a native support for handling of white space. In fact it's the only good reason for a conversion in my opinion. Nardog (talk) 05:49, 10 August 2018 (UTC)
 * Alright, thanks, the non-breaking non-space should work fine. Esszet (talk) 13:47, 11 August 2018 (UTC)

“Brasil”
You know that the IPA transcription for “Brasil” is [bɾaˈziw], right? I couldn’t change it because the template is protected. Ten Islands (talk) 15:31, 7 September 2018 (UTC)
 * The documentation isn't protected. Fixed. Nardog (talk) 02:04, 8 September 2018 (UTC)

IPA requests
Do we have a page, like Graphics Lab, for requesting IPA transcriptions? There's nothing mentioned on Help:IPA, and a query on WP:VPM came up with nothing concrete. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 22:20, 18 November 2019 (UTC)
 * Reference desk/Language is the most likely venue to post transcription requests. Alternatively you can tag a term with Pronunciation needed, which categorizes the article under Category:Articles needing pronunciation. Nardog (talk) 00:20, 19 November 2019 (UTC)
 * Thank you. I'll try the former; I see that the latter has a backlog of almost 1,000 pages. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 10:21, 19 November 2019 (UTC)

A discussion could use your input
A discussion at Wikipedia talk:WikiProject Linguistics could use your input. --Gonnym (talk) 11:39, 22 January 2020 (UTC)

Template-protected edit request on 22 April 2020
Please change the full protection to semi-protection. I want to edit the doc page. 14bauhr (talk) 18:13, 22 April 2020 (UTC)
 * Full-protection-shackle-no-text.svg Not done: is usually not required for edits to the documentation, categories, or interlanguage links of templates using a documentation subpage. Use the 'edit' link at the top of the green "Template documentation" box to edit the documentation subpage. Izno (talk) 22:45, 22 April 2020 (UTC)

If there's >1 argument, behave like Wiktionary's IPA template
In wikt:Template:IPA, the template parameters are:
 * 1 language code
 * 2, 3, 4 ... pronunciation(s)
 * qual N qualifier for pronunciation N
 * n N reference note for pronunciation N
 * 1 disable syllable counting
 * sort manual sort key

Because w:Template:IPA has only one parameter max, and Wiktionary's has two min, these are completely compatible.

Therefore, please merge them, to keep current behavior if 1 parameter and use Wiktionary's behavior if >1. This would help when copying content from Wiktionary. Sai ¿? ✍ 11:22, 15 March 2021 (UTC)

Include language tags
I was expecting the templates to produce HTML like  to tag the text as English language transcribed as IPA. Can you modify the template to emit lang tags? --Error (talk) 00:29, 1 May 2021 (UTC)

Small Text
Per MOS:SMALLTEXT it says Reduced or enlarged font sizes should be used sparingly, and are usually done with automated page elements such as headings, table headers, and standardized templates. Furthermore, it states that these styles should not be used in infoboxes and other templates that already reduce size. I note that a lot of IPA templates are used in the in the opening sentences and across infoboxes when introducing a non-English subject. Please can the automated formatting be removed to comply with the relevant MOS? >> Lil-unique1  (  talk  ) — 18:37, 15 July 2022 (UTC)
 * accessibility is a very important issue on Wikipedia, and I think these templates are not out of line with the MOS. The font size is governed by CSS styling and can therefore be adjusted by users/readers similar to how they can adjust the size of headers and text with their browsers. I think it's the in the span tag that governs that. So if you need a larger font, all you need to do is adjust the font size in your browser. You can also change to a different font if you like. Just read the  Usage  section on the template page. Hope this helps!  P.I. Ellsworth &thinsp;,  ed.  put'r there 20:09, 15 July 2022 (UTC)
 * thanks for the swift response. The MOS is equivocal and clear about the fact that small text being used sparingly. The use of small font in this template has no obvious purpose other than stylistically. Aside from aesthetics, it doesn't look good mid text to have small text. See Alesso for an example of what I mean. If there is a reason that IPA and its sub templates such as IPA-it render the resultant text in small font, small than standard prose, I'd like to know why otherwise this is a deviation of the MOS for no apparent reason. This arbitrary use of Small Text has started to be removed from other templates such as tracklist. The issue isn't overall text size, its that when you use this template, it generates some of the text in a smaller font for no reason. >> Lil-unique1  (  talk  ) — 20:14, 15 July 2022 (UTC)
 * Let's address that "no reason" idea. It appears that these and other similar templates have been rendering small fonts for many, many years now. It's good to question the reason, I think. So if there were good reason to reduce font size temporarily within article content as these templates do, what do you suppose that reason might be? Please indulge me on this. I think I know the reason; however, if you can come up with the reason on your own, then we might conclude that it is a logical and valid reason. Thank you for your patience and indulgence!  P.I. Ellsworth &thinsp;, ed.  put'r there 20:23, 15 July 2022 (UTC)
 * The only reason I could conclude that any text should ever appear smaller than the standard text size is because it is providing ancillary information, i.e. a note or some secondary information that is not pertinent to the immediate point. Arguably there could be some limited circumstances where this might be plausible such to ensure that a table renders on a standard screen size. However, it is unprofessional to have multiple text sizes within the same document or page. This is even alluded to in the MOS where it mentions that small text size via an appropriate template or CSS class should be limited in use. I myself would not be classified as visually impaired, nor do I require the use of screen reading or adjustment software. However, I see no valid reason for the template to render some information small and some not. In many instances of use, it is used within parenthesis which already indicate it is providing ancillary information. I can only fathom that the template was designed as such so that the smaller text is more like a note - however this would have been before Web Accessibility standards became more widely know. Taking accessibility out, from an aesthetic point of view I cannot see a scenario were it is pleasing or looks like deliberate/appropriate formatting. Where I have seen said IPA templates used, its nearly always in the middle of prose and therefore having random small words in the middle of a sentence looks and feels odd. I'm open to being convinced otherwise, but where there is no procedural reason based on functionality etc. I do think we should always aim for the maximum possible accessibility. In my example from the Alesso article, this means the template would not render the text Italian pronunciation in small. Prose underneath a section header should be in one size as this aids readability whether you have a disability or not. >> Lil-unique1  (  talk  ) — 23:23, 15 July 2022 (UTC)
 * Well, as someone who uses Chrome's largest font ("Very large") with an additional zoom (110%) just for comfortable reading and not because I'm visually impaired, I too can see no strong purpose for the small text other than what you mentioned...
 * Comparison:
 * Alessandro Renato Rodolfo Lindblad (born 7 July 1991),[1][2] better known by his stage name Alesso is a Swedish DJ and record producer.
 * Alessandro Renato Rodolfo Lindblad (born 7 July 1991),[1][2] better known by his stage name Alesso (Italian pronunciation: [aˈlɛsso]) is a Swedish DJ and record producer.
 * I really have no strong opinion either way. I think that increasing the font size for this template would mean changing the to something else. That would then have a domino effect for  several other applications . So it appears that a  consensus  would be needed, and that means finding a centralized place to  discuss it  such as WT:LANG or WT:IPA or even WT:ACCESS. Not sure which venue would provide the best turnout of editors; however, placing neutral notices of where the discussion is being held can be very helpful.  P.I. Ellsworth &thinsp;,  ed.  put'r there 16:15, 16 July 2022 (UTC)
 * Quick note: The  class has nothing to do with the label size. That code is in IPA-xx templates, not IPA. Nardog (talk) 17:58, 16 July 2022 (UTC)
 * Thank you for that! So you're saying that each difflang IPA template would have to be changed? And that is so even if consensus were to emerge to better meet accessibility and aesthetic issues by increasing the font size? Well, there aren't that many of them, maybe 150 or so? And what exactly does the IPA class in this template do?  P.I. Ellsworth &thinsp;, ed.  put'r there 18:26, 16 July 2022 (UTC)
 * See the template documentation. It used to be assigned a font stack in MediaWiki:Common.css but not anymore because most devices can show IPA characters by default now. Nardog (talk) 23:50, 16 July 2022 (UTC)
 * That's why I thought the class=IPA would determine the font size as well as the font type. Kind of odd that the small tags would be used in the individual IPA templates. Guess no one thought it would ever be questioned? and it was probably thought that not all the text should be rendered small, just a part of it.  P.I. Ellsworth &thinsp;, ed.  put'r there 18:14, 17 July 2022 (UTC)

, you have fixed several MOS:SMALLTEXT issues before. I came here after seeing the infobox in Isa (name) and having to squint to read it. There is discussion above about needing consensus to change anything. But MOS:SMALLTEXT is clear on this. The real issue is that these templates are commonly used in prose where there is no problem (e.g. Najd). So isn't the solution that we need a 100% parameter as in post-nominals so the default of small can be overwritten only when needed, as in infoboxes?

For reference, the doc at that template says Font size: The optional That is exactly the same as here. It even says must; I think the small text must be fixed here also. MB 00:40, 30 September 2022 (UTC)
 * The size of the word "Arabic" in Isa (name) is not in compliance with MOS:FONTSIZE, because a font size of 85% is being applied to text that is already reduced to 88% of nominal by the infobox. The challenge is that the 85% size is being applied by  tags in IPA-ar and its siblings, not by IPA. This IPA template applies only to the rendered pronunciation characters; as far as I can tell,  is applied only to those characters, and there is no size reduction here. There may be a clever way around this double-small problem; I'll take a look. I think the grand solution is to have all of this rendering done by a single template (or module, presumably) instead of by nearly 200 essentially identical templates (edited to add: and 130+ more). That would involve a bit of programming by one of two or three editors I can think or whom I will not name yet, lest I incur their annoyance. – Jonesey95 (talk) 05:21, 30 September 2022 (UTC)
 * @Jonesey95, yeah, I was aware that a solution will impact all those 200 templates in one way or another. After something is worked out, there is also the matter of editing all the articles that use these templates in an infobox. Probably a job for a bot. For now, I'll leave this in your capable hands, but let me know if there is anything I can do to help. MB 05:33, 30 September 2022 (UTC)
 * See also Wikipedia talk:WikiProject Linguistics/Archive 16 and Wikipedia talk:WikiProject Linguistics/Archive 17 for users who have expressed interest in overhauling these templates. I've already written modules to improve and replace these templates years ago, but a switch would involve consensus building and extensive bot work as the templates have developed inconsistencies.
 * As for the small text, I'm of the opinion that we should probably just do away with it, or at least make it an opt-in, to align with lang-xx templates. Nardog (talk) 12:27, 30 September 2022 (UTC)
 * It is good news that the modules are already (mostly?) developed! I think that the next step may be to do extensive test cases and then a proof of concept with one of the easier IPA-xx templates, converting it to a wrapper of #invoke:IPA or whatever will work. We may not need a bot for that step. If the proof of concept is shown to work, we can then do a TFD, at least for the templates that appear to be straightforward to convert. We can convert the templates to wrappers one-by-one, and then if it is shown to work, a bot can follow up and convert each IPA-xx to IPA|lang=xx or whatever the new syntax will be. Templates that are more difficult to convert can be done in a second phase. Let's get the ball rolling. – Jonesey95 (talk) 13:23, 30 September 2022 (UTC)