Module talk:ca-IPA

Double stress

Compound words and adverbs with -ment are doble stressed with a secondary and a primary stressed vowel. For example, sobreviure is /ˌso.bɾəˈbiw.ɾə/ not /su.bɾəˈbiw.ɾə/, and bojament is /ˌbɔ.ʒəˈmen/ not /bu.ʒəˈmen/. I think it can be handled as two separate words concatenating them afterwards, maybe with some problems with contact assimilations. This should be passed as a parameter indicating somehow that the word is fomed by two parts. Another option is to pass the word with two accents, sóbrevíure and bòjamént. In any case the structure of the module is affected. Any thoughts how to handle that? --Vriullop (talk) 19:25, 1 September 2016 (UTC)

Gone ahead passing the word with two accents. --Vriullop (talk) 12:29, 11 September 2016 (UTC)

ê and ô

These notations aren't still supported when a full word is overwritten in the template. How can this be fixed? [ˌiˑvã̠n̪ˑˈs̪kr̺ud͡ʒʔˌn̺ovã̠n̪ˑˈt̪ɔ̟t̪ːo] _{(parla con me)} 16:45, 15 October 2016 (UTC)

Fixed. Now you can use ê/ô either as hint for the pagename or included in the word as argument: {{ca-IPA|Francêsc}} --Vriullop (talk) 19:40, 15 October 2016 (UTC)

• (Balearic) IPA^(key): /iɾˈlan.dɛs/
• (Central) IPA^(key): /irˈlan.dɛs/
• (Valencian) IPA^(key): /iɾˈlan.dɛs/

What's happening in irlandès? I type {{ca-IPA|irlandês}}, but it gives the result shown at the right (using Special:ExpandTemplates). It seems to be transcribing the page name, irlandès, but with penultimate rather than final accent, rather than the word that I put in the first parameter, irlandês. — Eru·tuon 06:28, 20 January 2017 (UTC)

Fixed. It works with {{ca-IPA}} assuming all -ès endings are ê case, also with {{ca-IPA|ê}} changing the stressed vowel, and now with {{ca-IPA|irlandês}} detecting ê as stressed vowel. --Vriullop (talk) 08:14, 20 January 2017 (UTC)

@Vriullop I think that's a bit misleading. The spelling irlandès would never be pronounced with /e/, that would be irlandés. There should be no Valencian pronunciation listed at all. —CodeCat 21:46, 23 January 2017 (UTC)

@CodeCat In Valencian both forms are used, see the entry in Diccionari Normatiu Valencià, althought both pronounced /e/. Forms in -és are the preferred ones in Valencian standard but historically -ès forms were used and nowadays some Valencian writers still prefer general standard form. Moreover, in North-western Catalan spelling is always irlandès pronounced /e/. --Vriullop (talk) 09:02, 24 January 2017 (UTC)

Weird. You'd think, with the regular spelling rules that Catalan has, people would write what they say. —CodeCat 14:25, 24 January 2017 (UTC)

Not so easy, there are a lot of local variations. A good example is València that most Valencians pronounce /e/. Eastern Catalan people have difficulties for writing unstressed o/u or a/e while Western Catalans with é/è. The trick is to think how people on the other side pronounce it. --Vriullop (talk) 17:37, 24 January 2017 (UTC)

Identical IPAs

I was wondering whether it would be possible to resume different dialect pronunciations in a single line, when these are identical; for instance:

• (Balearic, Central) IPA^(key): /əˈma/
• (Valencia) IPA^(key): /aˈmaɾ/

Instead of:

• (Balearic) IPA^(key): /əˈma/
• (Central) IPA^(key): /əˈma/
• (Valencia) IPA^(key): /aˈmaɾ/

Thanks! [ˌiˑvã̠n̪ˑˈs̪kr̺ud͡ʒʔˌn̺ovã̠n̪ˑˈt̪ɔ̟t̪ːo] _{(parla con me)} 18:05, 23 January 2017 (UTC)

That is a great idea. I tried to make the module do it, but it doesn't work for some reason. — Eru·tuon 01:46, 24 January 2017 (UTC)

Oops, just added the case where Balearic = Central. Now it works! There might be a simpler way to do it, I don't know. 01:49, 24 January 2017 (UTC)

@Erutuon: I've written a function that does it agnostically to the specific accents (so if an accent were added later, nothing would need to be changed). This may be a bit of overkill, so, if you wanna' revert it, go ahead. —JohnC5 03:02, 24 January 2017 (UTC)

@Erutuon, JohnC5: well, thank you a lot in any case! [ˌiˑvã̠n̪ˑˈs̪kr̺ud͡ʒʔˌn̺ovã̠n̪ˑˈt̪ɔ̟t̪ːo] _{(parla con me)} 13:18, 24 January 2017 (UTC)

març

Why does it delete the /ɾ/ in Balearic and Central? (Also is anyone working on phonetic transcription?) Ultimateria (talk) 14:52, 5 February 2017 (UTC)

Fixed no silent r in -rç endings, it is silent only in some -rs endings, most plurals of silent -r. --Vriullop (talk) 11:30, 6 February 2017 (UTC)

caràcter

ca.wikt has the final r pronounced in Central, is that right? Ultimateria (talk) 18:27, 13 February 2017 (UTC)

@Ultimateria This is one of fluctuating cases pronounced in formal and emphasized speech and silent in informal or relaxed speech. You can use -rr hint {{ca-IPA|caràcterr}} for final r formally pronounced in Central Catalan or expand it with an explanation. --Vriullop (talk) 10:23, 18 February 2017 (UTC)

Coda -lls

@Erutuon About Special:Diff/46172315, it sounds to me that /ʎʃ/ is uncommon. In DIEC there are 30 results for -lls, apart from other plurals of -ll. In Diccionari Ortogràfic i de Pronúncia and Diccionari Valencià de Pronunciació all these words are annotated with /ʎs/. Searching for any dialectal variation I have only found one word registered with /ʎʃ/ in some villages corresponding to w:ca:Català pallarès, a dialect of Northwestern Catalan. Some blogs report /ʎʃ/ also in some Valencian locations, but it is not general at all. --Vriullop (talk) 18:19, 11 June 2017 (UTC)

@Vriullop: Okay, I wonder why w:Catalan phonology says -lls is pronounced /ʎʃ/ in Eastern Catalan and Valencian, but the sources you cite do not.

I made this edit because I remember hearing /ʎʃ/ for -lls in a Catalan soundfile. I can't remember where it was but see for instance this sound file of ells by a speaker from the Balearic Islands and this of cavalls from Catalonia. It would puzzle me if -nys were pronounced /ɲʃ/ (as is already noted by the module) but -lls were not similarly assimilated. I suppose such irregularities frequently happen in languages, though. — Eru·tuon 18:49, 11 June 2017 (UTC)

@Erutuon: I have to admit I am quite confused. It seems not so uncommon as I said at first glance. First file of ells does not sound strange to me, I can recognize it as Balearic. File of cavalls uses too much emphasis, it is no natural speech and I would not consider it. See the linguistic atlas for anells, budells, cabells, grills, ulls. It alternates [ʎs]-[ʎs̠]-[ʎʃ]-[ʎtʃ]. The last one is apitxat Valencian, the three first ones alternates without a clear dialectal pattern. There is also a [js] but it is another feature. I have discussed it with some colleagues and I would say that, appart from local tendencies, /ʎs/ is regular pronunciation and /ʎʃ/ is relaxed speech. On the contrary, it seems that /ɲʃ/ is more common than /ɲs/ althought there is also a range of variations. Summarizing, I think it is better to annotate it as an alternative pronunciation in entries, as in Special:Diff/46179010. Currently the module is not prepared to introduce alternatives. --Vriullop (talk) 09:04, 12 June 2017 (UTC)

ll in Balearic

@Vriullop: In certain words, such as mirall, Balearic has /j/ instead of /ʎ/ (this is whenever the "ll" comes from older /jl/ instead of /ll/). We should add a hint to indicate that. --WikiTiki89 16:00, 14 March 2018 (UTC)

@Wikitiki89 It can be pronounced both /ʎ/ and /j/ depending of the context formal or informal. A TV presenter will pronounce /ʎ/ but /j/ in a colloquial conversation. In ca:mirall I noted it as informal, as it is done usually in Catalan papers. Template {{ca-IPA}} is not prepared for including informal cases, or we should discuss if it should include the formal pronountiation or the most common one.

For more information you can see "iodització" on Diccionari de Dubtes del Català Oral ("a Mallorcan feature in informal register") or Atles Lingüístic del Domini Català ("an unestable phenonemon in regression"). Note that it can be /j/ also in Central Catalan, but it is underestimated as too relaxed speech associated with Spanish yeísmo. --Vriullop (talk) 17:48, 14 March 2018 (UTC)

@Vriullop: Just to be clear, will someone who speaks Balearic pronounce res as [ˈrɛs] or [ˈrəs] when speaking formally? Because it seems to me that Balearic accent itself is informal. --WikiTiki89 18:03, 14 March 2018 (UTC)

@Wikitiki89 It is pronounced /ə/ formally in Mallorca and /ɛ/ in most Ibiza and Menorca. Maybe a literate Mallorcan changes to /ɛ/, but this is a deeply rooted feature not perceived as informal and not so easy to change in different registers. --Vriullop (talk) 19:09, 14 March 2018 (UTC)

@Vriullop: Got it. So then I think we have to somehow be able to support the informal Balearic /j/ in place of /ʎ/. But I guess that would be a more complicated change than I originally expected. So no pressure to do it anytime soon, but it would be nice to have. For now, I'll add this pronunciation manually to entries (like this). --WikiTiki89 19:19, 14 March 2018 (UTC)

Porting and re-licensing module code

I'm currently in the process (and almost done) porting this module to JavaScript and making it into an NPM module (that can also run standalone in a browser environment). The problem is that module code on Wiktionary (or any other MediaWiki project using the Scribunto extension, as far as I have seen) is not treated differently to any other content, meaning that it's also subject to the CC BY-SA 3.0 license. However, Creative Commons licenses are not recommended for software/code, so I would like to use another license if possible. In answer I got over on opensource.stackexchange.com, it has been suggested that I would have to modify the code to such an extent that I could relicense it as CC BY-SA 4.0 and then I'd have to modify it again (to a similar extent) in order to be able to publish it under GPLv3 (which is regarded as compatible to CC BY-SA 4.0.

This is a general issue I'd like to bring up with a broader audience over at w:en:Wikipedia_talk:Lua at some later time, but I'd like to find a solution for this specific case first, as I'm itching to get this package published and don't expect a general solution in a timely manner.

Now I don't think porting to another language in itself is regarded as a big enough change (doesn't really include any creative work) to enable relicensing, and even the adaptations for converting it into an NPM module, stripping away any unncecessary Wiki environment related stuff, are most probably not enough. That's why I'd like to ask you, @Vriullop, @JohnC5 and @Erutuon, as the only authors (with the exception of one minor edit in a debug message) of this module for your consent to publish your work under a different license. I would personally prefer the MIT license, but being able to directly relicense under GPLv3 would be fine as well, I guess. It would be great if the three of you could state your general thoughts and the license(s) you'd be willing to allow.

I will of course still give attribution to you for your great work in the README of the package. I'd also like to port back any fixes or extensions that might come in through the package repo, given that my time and lousy knowledge of Lua allows. And just to make sure: Of course this agreement, should we reach one, shall only apply to the ca-IPA module code (I don't use any of the required modules in my port).

Thank you for your time, looking forward to your feedback. --Connum (talk) 08:15, 4 July 2019 (UTC)

@Rua pinging another major author. --Vriullop (talk) 13:26, 4 July 2019 (UTC)

Thank you! I should have expanded the complete history... ;-) Also pinging @IvanScrooge98 --Connum (talk) 15:31, 4 July 2019 (UTC)

Well, I did veeeery little on this module, and don’t have much knowledge on licenses either, so I’ll just let the major authors intervene. Thanks for pinging me though. [ˌiˑvã̠n̪ˑˈs̪kr̺ud͡ʒʔˌn̺ovã̠n̪ˑˈt̪ɔ̟t̪ːo] _{(parla con me)} 15:45, 4 July 2019 (UTC)

@Connum I have no problem in relicensing my contributions to another free license provided that the attribution of the authors and the project is maintained. Therefore for me both the MIT license and the GPLv3 are fine. --Vriullop (talk) 10:03, 8 July 2019 (UTC)

Thanks for your feedback, that's great! --Connum (talk) 11:27, 8 July 2019 (UTC)

@Connum: I also have no problem with it, under the same conditions mentioned by Vriullop. — Eru·tuon 16:14, 9 July 2019 (UTC)

Thank you! Two to go... :-) From the feedback so far, it looks like it's going to end up as BSD-3-Clause, as that can be regarded as identical to MIT, except having a stronger enforcement of attribution (not only in source code but also in binary distributions). This seems to fit everyone's interests best (unless @JohnC5 or @Rua are going to raise objections). Cheers! --Connum (talk) 20:49, 9 July 2019 (UTC)

I would prefer to licence my code under a copyleft licence, so that others can continue to benefit. —Rua (mew) 20:54, 9 July 2019 (UTC)

thanks for your feedback! Would LGPL work for you, or do you require a stronger copyleft (in which case I would use GPLv3)? --Connum (talk) 07:21, 10 July 2019 (UTC)

LGPL would be fine, although I'm not sure where the difference would actually matter. —Rua (mew) 09:01, 12 July 2019 (UTC)

Thanks again. I'm not that deep into license terms either, but as far as I understand, LGPL is better for libraries. I just checked the contributions and found out that the part that @JohnC5 added to the code (the grouping functionality) will currently not end up in my port, so now I have the OK of all contributors, with the most strict license demanded being (L)GPL. I'll still wait for another couple of days to allow for any further input before I'm going to publish the package to npm. Thanks again to all of you! --Connum (talk) 08:03, 15 July 2019 (UTC)

For what it's worth, I don't mind anyway. —*i̯óh₁n̥C^[5] 09:10, 15 July 2019 (UTC)

Thanks for your contribution and for getting back to me, anyway! :) --Connum (talk) 10:16, 15 July 2019 (UTC)

The package is now available on NPM and as a GitHub repository. Thanks again to everyone, for the contributions as well as the straightforward clarification of the re-licensing issue. --Connum (talk) 22:58, 18 July 2019 (UTC)

Alternative pronunciations

Looks like this is currently missing. ca:Template:ca-pron supports this with |or=, |or2= etc. – Jberkel 18:30, 1 December 2019 (UTC)

Most relevant alternative features missing are:

• For Valencian v is pronounced /v/ in Central Valencian but /b/ in other areas.
• For Balearic unstressed o is pronounced /o/ in most Majorcan but /u/ in some Majorcan areas and in the two other islands.

--Vriullop (talk) 21:09, 1 December 2019 (UTC)

[ ] versus / /

@Benwing2 The nasal assimilations go beyond the phonemic level as there is no */ɱ/ for example. At that point one might as well switch to phonetic transcriptions with [β ð ɣ] and so forth. Nicodene (talk) 21:31, 26 January 2024 (UTC)

@Nicodene Indeed, that is the plan, as I think it's misleading not to display the approximants. User:Vriullop has argued that we should stick with /.../ notation even when displaying the major allophones; maybe they can comment on why. Benwing2 (talk) 21:42, 26 January 2024 (UTC)

I see. I would support that change. Perhaps Vriullop is thinking of Catalan Wiktionary, which does put allophones in / /, as in /iɱ.fiˈðɛɫ/ for infidel. There is no need to follow others' misuse of linguistic notation, however, and if cross-wiktionary consistency is desired it can be achieved by contacting their editors about this oversight. Nicodene (talk) 22:24, 26 January 2024 (UTC)

@Nicodene I have implemented [β ð ɣ] and switched the slashes to brackets. Benwing2 (talk) 02:56, 27 January 2024 (UTC)

Nice. It'd always bothered me a bit to only have a representation /b/ (even if phonologically sound) in cases where there's never even been a [b], like divendres < Latin dies veneris. Nicodene (talk) 08:44, 27 January 2024 (UTC)

Elision of contiguous vowels

@Benwing2 Words like sobreestimar, guardaespatlles or antiespasmòdic should loss the first vowel of the second component in Central and Balearic. It is done in the Catalan version, see function elisionContinguousVowels, but it's with the old code with table of syllables, not sure how to implement it here. It is described in GIEC 3.4 but in a rather confusing way that I have summarized in comments at the module. Vriullop (talk) 18:23, 7 March 2024 (UTC)

@Vriullop OK thanks. I will take a look at that function and see about duplicating its behavior here. Benwing2 (talk) 06:30, 8 March 2024 (UTC)

BTW FIXME's #5/6/7 in the module comments refer to elision of adjacent vowels. I didn't implement it before mainly because I wasn't sure of the rules. Benwing2 (talk) 06:31, 8 March 2024 (UTC)

@Benwing2 After reading your comments, I have to review the <oe> case. It is always compressed followed by s. With other coda consonants, o without coda, I can not see any pattern to follow: fotoelèctric is compressed but not magnetoelèctric, radioenllaç is compressed but not radioemissora nor autoencesa. It will be safer to include additional cases other than <oes> by respelling. Vriullop (talk) 10:55, 8 March 2024 (UTC)

Unknown pronunciations

@Benwing2: Whenever one of the dialects is set to ?, e.g. at guèiser, the page gets put into Category:IPA pronunciations with invalid representation marks. I'm not sure exactly why; could you take a look and fix it? Thanks! —Mahāgaja · talk 12:27, 20 May 2024 (UTC)

@Benwing2: Did you see this? —Mahāgaja · talk 07:28, 29 May 2024 (UTC)

@Mahagaja Should be fixed. Benwing2 (talk) 08:04, 29 May 2024 (UTC)

/ld/

As has been pointed out here, there’s a mistake in the code below line 957. /ld/ does not phonetically lenite to [lð], contrarily to what /ɾd/ and other sequences do. Can anyone fix this? [ˌiˑvã̠n̪ˑˈs̪kr̺ud͡ʒʔˌn̺ovã̠n̪ˑˈt̪ɔ̟t̪ːo] _{(parla con me)} 22:56, 16 November 2024 (UTC)

Done. It was noted in the comments, but not applied. --Vriullop (talk) 11:50, 19 November 2024 (UTC)

And since we're at it, it seems there are a few recent borrowings where, at least from the audiofile, it seems "e" doesn't reduce to [ə] like in làtex and píxel; I tried and I tried but there doesn't seem to be a way to have that unstressed e appear as [e] without the module automatically introducing secondary stress. Sérgio Santos (talk) 00:08, 17 November 2024 (UTC)

In formal speech, both 'làtex' and 'píxel' should reduce the unstressed vowel. But you're right, some recent borrowings may not be fully adapted, or be hesitant depending on the speaker. Also, for other reasons, in Balearic there is no reduction in a handful of words. For example, pegar is pronounced [peˈɣa] avoiding homophony with pagar. In cawikt this is noted with respelling ē or ō. Pinging Benwing2. Vriullop (talk) 12:14, 19 November 2024 (UTC)