Module talk:es-pronunc

See also: Module:es-pronunc/testcases

gü doesnt work correctly yet

Template:es-pronunc, correct pronunciation at cigüeña.Matthias Buchmeier (talk) 15:36, 9 September 2014 (UTC)

This problem has been solved (dunno when or by whom), the output is now correct (diff). --Per utramque cavernam (talk) 21:10, 17 April 2018 (UTC)

Currently I see /θiɡˈweɲa/, [θiɣˈweɲa], /siɡˈweɲa/, [siɣˈweɲa]. I think it should be /θiˈɡweɲa/, [θiˈɣweɲa], /siˈweɲa/, [siˈweɲa]. Copihue and chirigüe are perfect rhymes. Huaso and guaso are variant spellings, not variant pronunciations. Piaractus (talk) 15:36, 14 October 2019 (UTC)

@Piaractus: Fixed syllabification in cigüeña. I'm not confident enough about the details of Spanish pronunciation to make the other change you mention. — Eru·tuon 16:42, 14 October 2019 (UTC)

Are g's in this context dropped in all varieties of Latin American Spanish or only some? —Granger (talk · contribs) 12:19, 15 October 2019 (UTC)

Phonemic transcription

Not long ago, @Mr KEBAB has removed the diacritics on phonemic symbols and undone my edits on the module, which I had made to have a close transcription, as in Module:ja-pron or Module:ko-pron, for example. He justified this saying they were useless, but can we actually consider diacritics useless in phonemic IPA? After all, the phonetic one (not needing diacritics) is given just before it, and the phonemic one is intended to represent a more precise pronunciation. So, does anyone agree with me this should be reverted, at least partially? IvanScroogeNovantotto _{(parla con me)} 09:10, 23 July 2016 (UTC)

@IvanScrooge98 Sorry for the delay - that's the cost of wanting a long message to be well-sourced :(

I removed them partially because they made the narrow transcription biased (as far as different descriptions of Spanish pronunciation are concerned) and, in some cases, Eurocentric (biased towards European Spanish), and partially because there are levels of narrowness of phonetic transcriptions, which is very important to understand. (And you probably meant "in phonetic IPA", since that's what we're discussing).

We don't need all to use all of the diacritics that are available. In fact, practically no scholar does, maybe save for a small amount of researchers like Hans Basbøll in his impressive "Phonology of Danish" (2005). But even he uses them exclusively when he finds it appropriate, and a verbal explanation always follows. I think that my efforts to simplify the set of symbols we use were reasonable, but I'll try to explain them anyway.

I'm aware that our Spanish phonology article on Wikipedia is less than perfect (as far as details are concerned), but I'll try to change that. The same goes for Appendix:Spanish pronunciation.

First of all:

The denti-alveolar stops /t, d/ are correctly transcribed [t, d] or [t̪, d̪], because the signs [t, d] do not denote alveolar sounds by default - they can be dental or alveolar, depending on the language (see Handbook of the IPA (1999:8, 164-165)). Besides, [t̪, d̪] is not a completely narrow transcription either, because they can denote denti-alveolar or true dental sounds, which makes that set no better than [t, d] at all. However, since we're transcribing the denti-alveolar allophone of /n/ as [n̪], it made some sense to bring back the notation [t̪, d̪], and that's what I did.

I'm now going to explain most of the changes I've made to this module:

Vowels

- There's no agreement among scholars that the mid vowels are true-mid in all positions; some of them claim that there are close [e, o] and open [ɛ, ɔ] allophones (see e.g. this source). The source I've just linked to also claims that there are close [i, u] and open [i̞, u̞] allophones of the close vowels /i, u/, whereas /a/ has three allophones - front [a], central [a̠] and back [ɑ]. Yet there clearly are scholars (e.g. Martínez Celdrán et al. (2003:256)) who claim that there are no such allophones (see also the discussion in Nowikow's "Fonetyka hiszpańska" (2012:19-20), kindly translated by a Wikipedian). Transcribing /e, o, a/ as if they had only one allophone [e̞, o̞, a̠] is clearly biased towards the latter view, and that's something we shouldn't be doing.

- There are no separate symbols for the true-mid vowels and the open central unrounded vowel in the IPA, and [e, o, a] are the preferrable symbols in this case (see Handbook of the IPA (1999:9, 13), and also these blogposts ([1], [2], [3]) by John C. Wells, a highly respected British phonetician), it's clearly a correct-but-not-fully-narrow phonetic transcription even if you subscribe to Martínez Celdrán's point of view (which I don't find to be acceptable - we should remain neutral whenever possible).

Voiced stops /b, d, ɟ͡ʝ, ɡ/, including their continuant allophones

- The sounds [β, ð, ʝ, ɣ] are not considered approximants by all authors (see [4]), so the broader transcription [β, ð, ʝ, ɣ] is a suitable narrow notation - all we're doing is dropping the undertack. The undertack-less notation is also endorsed by the IPA (see Handbook of the IPA (1999:9)), who consider both transcriptions to be equally correct. By the way, to see why [j, ɰ] is not a correct transcription of the last two sounds (not least because [j] is used for a different sound in Spanish IPA, the one that can be alternatively transcribed [i̯]), see Martínez Celdrán (2004).

- As far as the symbols [ʝ, ɣ] are concerned, an undertack below them may not be displayed properly in all browsers, so some (or many?) users can end up with [β̞, ð̞] being displayed correctly, while [ʝ̞, ɣ̞] are not. Putting the "lowered" diacritic to the right ([ʝ˕, ɣ˕]) looks very bad when other sounds follow (see for yourself: [aˈʝ˕uð̞a], [ˈalɣ˕o] - it looks as if [˕] were a sound in and of itself), so it shouldn't be done unless there's a very good reason to do so.

- The /ɟ͡ʝ/ phoneme (which can also be symbolized as /ɟ/ or /ʝ/, although I'm not a fan of the last transcription, as it strikes me as inconsistent with the other stops) in post-pausal word-initial position varies between a continuant [ʝ] and an affricate [ɟ͡ʝ], with the latter variant occurring in careful speech (Nowikow's "Fonetyka hiszpańska" (2012:41)). I'm not making a plosive-affricate distinction here, since some authors (e.g. Nowikow) consider it to be a plosive [ɟ], yet others describe it as an actual affricate. I'm not sure whether we should allow transcribing the post-pausal word-initial [ʝ] along with the more careful affricate variant, but if we do, we need to make sure that when a word-final nasal precedes, only transcription with [ɟ͡ʝ] shows up.

Voiceless dental fricative /θ/

- I see no reason to transcribe /θ/ as [θ̟]; this is possibly, again, a biased transcription - see [5], where the quote says "(...) which, as we shall see, brings it quite close to one variety of /θ/ (...)", which clearly implies that there are more than one possible realizations of that phoneme, and we're not sure what they are. Maybe it's a mistaken interference from [n̟, l̟]? The symbols [n̟, l̟] (the convention of Martínez Celdrán et al. (2003)) have an "advanced" diacritic underneath them to denote their interdentalness because [n, l] are rarely interdental in world's languages (and, again - [n, l] are not by default alveolar in IPA, they denote dental or alveolar sounds, depending on the language (see Handbook of the IPA (1999:8, 164)). [θ], on the other hand, is defined as dental by the IPA (see Handbook of the IPA (1999:170)), so no diacritic is needed here. Besides, even Martínez Celdrán et al. (2003) themselves do not transcribe [θ] as [θ̟], just [θ] (without the "advanced" diacritic), even though they call it "interdental" (as they call the approximant [ð]).

- I changed the phonetic transcription of the voiced allophone of /θ/ from [ð] (now used for the continuant allophone of /d/) to [θ̬]. It is not an OR transcription, some scholars use it (e.g. Nowikow in his "Fonetyka hiszpańska" (2012)). It is also not really wrong, because the "voicing" diacritic doesn't tell you how voiced the sound in question is, just as the "devoicing" diacritic doesn't tell you how devoiced certain sound is (see Handbook of the IPA (1999:15-16)). It's questionable if we should use that symbol at all, because, in case of /θ/ there appears to be a free variation between voiced [θ̬] and voiceless [θ] in environments in which other voiceless fricatives (/f, s/ but not /x/) are voiced (see [6]).

Voiceless alveolar sibilant /s/

- I see no reason to transcribe /s/ before the dental stops as dental [s̪], as this introduces discrepancy as far as the Latin American transcription is concerned. The existence of this allophone of an otherwise alveolar /s/ has been reported only for the Standard European accent (Martínez Celdrán et al. (2003:258)), and many Latin American speakers use the dental [s̪] as their ordinary /s/ (see Nowikow's "Fonetyka hiszpańska" (2012:50)), not just before the dental stops. It is best to just transcribe it [s], it's a neutral symbol.

- It's completely wrong to regard the aspirated allophone of /s/ to be a phoneme /h/, because it's not. Also, /s/-aspiration is non-standard as far as Mexico and Colombia (especially the latter is commonly regarded as one of the clearest accents of Spanish) are concerned, so I think we shouldn't list it at all.

Voiceless velar fricative /x/

- The phoneme transcribed /x/ varies phonetically between velar, post-velar, uvular and glottal, depending on the dialect. There have been reports of European speakers using the uvular realization as their normal one (see [7]), and I've never seen one source that said it's non-standard, so I'm fairly confident that we shouldn't use any other symbol than [x] in our transcription, because that is also a correct broad transcription of the voiceless uvular fricative [χ] (see [8]). Also, according to Martínez Celdrán et al. (2003:258), the uvular allophone is merely a possibility, yet according to Nowikow's "Fonetyka hiszpańska" (2012:45), there's no such free variation, and the uvular allophone seems to be described as the only pronunciation possible (or at least "recommended"), so there's no agreement here as well.

Raised labialized velar approximant /w̝/

- Wikipedia doesn't say how standard it is (see [9]), and it's probably not a part of the Standard European accent at all. It's best to omit it.

Other allophones

- There are also other allophones in Spanish, such as the post-palatals [k̟, ɡ˖, ɣ˖, x̟], which you chose not to cover here. I agree with that, it is also controversial whether [x̟] exists at all for speakers with a uvular [χ] in all positions... besides, we'd have the same problems with the symbols [ɡ˖, ɣ˖] (or [ɡ̟, ɣ̟]) as with "diacricized" ordinary [ʝ, ɣ] (see above), so that's another problem.

I hope that my message covered everything you wanted to know. Sometimes it's best to keep things simple, or at least as reasonably simple as possible. This also applies to IPA. Mr KEBAB (talk) 23:19, 25 July 2016 (UTC)

@Mr KEBAB: thank you so much for replying; wow, you made a very detailed research, and so well-referenced I can’t argue any point of it! Great job, and again thanks for being so accurate and stopping me from making other edits!! BTW, I meant phonetic with phonemic and vice versa, another mistake. IvanScroogeNovantotto _{(parla con me)} 09:24, 26 July 2016 (UTC)

@IvanScrooge98 Glad to hear that :) Mr KEBAB (talk) 11:47, 26 July 2016 (UTC)

Serious mistake above: what I meant was "I'm not sure whether we should allow transcribing the post-pausal word-initial [ʝ]" - the approximant, not the affricate. Mr KEBAB (talk) 12:45, 26 July 2016 (UTC)

productividad

Is the c seriously not /k/?? Ultimateria (talk) 21:15, 26 January 2017 (UTC)
And the p in septiembre is not /b/, is it?? @IvanScrooge98, Mr KEBAB. Ultimateria (talk) 15:09, 5 February 2017 (UTC)

@Ultimateria See [10]. Remember that /ɡ, b/ do not represent phonetic [ɡ, b] but [ɣ, β], which are the corresponding approximants or fricatives. I assume that you're aware of the difference between phonemes and allophones. Mr KEBAB (talk) 15:29, 5 February 2017 (UTC)

I suppose, but could it not be /k, p/ -> [ɣ, β]? Ultimateria (talk) 18:51, 5 February 2017 (UTC)

/k, p/ are typically not said to have the allophones [ɣ, β], but /ɡ, b/ are. If you're familiar with Cockney English, a related example is the vocalized /əl/ ('syllabic dark l') in words like awful, which native speakers of Cockney are ready to transcribe /ˈoːfoː/ (or /ˈoʊfoʊ/ - same phoneme, different notation). Also remember that orthography and phonetics/phonology are quite different things. Mr KEBAB (talk) 05:13, 6 February 2017 (UTC)

Comparison with existing IPA: User:DTLHS/Spanish pronunciation/phonemic

(moved from my talk page) @Metaknowledge Let's start off as simple as possible. Here I have generated a report showing Spanish entries with only phonemic pronunciations (and no accent qualifiers or other templates) that differ from the module output. You should bear in mind that I know very little about IPA so I'm going to need a lot of buy in from you to make this happen. DTLHS (talk) 04:14, 17 February 2017 (UTC)

I'm happy to help, and I'm sure others are as well, though I don't know whom to ping. The immediate issues causing mismatch that I see are that the module is not giving syllable breaks, which I think it should, and that it is using /b/ and /d/ instead of /β/ and /ð/ (technically equivalent choices, but I think we would be better off using the most specific IPA symbols we can without needing diacritics; others may disagree). —Μετάknowledge^{discuss/deeds} 06:36, 17 February 2017 (UTC)

@Mr KEBAB What do you think about these suggestions? DTLHS (talk) 22:20, 17 February 2017 (UTC)

@DTLHS We can't differentiate /b, d, ɟ͡ʝ, ɡ/ from */β, ð, ʝ, ɣ/ in phonemic transcription. The latter do not exist in Spanish as separate phonemes, they're just positional allophones of the stops. Transcriptions such as */diˈðaktiko/ are not phonemic, they're phonetic. I also suggest that you read my lengthy message above (or, better, the whole discussion), many of the issues with your transcriptions were already discussed there. Ping me if you'll have trouble with understanding it (it could be somewhat challenging). Mr KEBAB (talk) 17:33, 18 February 2017 (UTC)

flauta

Why does it start with /f/ but then [v]? Ultimateria (talk) 13:39, 18 March 2017 (UTC)

@Ultimateria I don't know how natives pronounce that word, but it's possible that word-initial /f/'s aren't voiced before sonorants. Mr KEBAB (talk) 15:52, 6 April 2017 (UTC)

@Ultimateria On second though that's almost definitely [f], not [v]. Mr KEBAB (talk) 02:25, 16 March 2018 (UTC)

Must introduce word-initial exception to

word = mw.ustring.gsub(word,'f([ˈˌ]?[mnɲbdɟɡlʎr])','v%1')

If there are words in which fl is pronounced [vl], they need to be added to the testcases before this rule can be changed. — Eru·tuon 20:16, 18 April 2018 (UTC)

I don't think there are any words like that. I can't think of any, and in pantufla (where it's after a vowel), for example, the pronunciation is [fl]. —Granger (talk · contribs) 20:23, 18 April 2018 (UTC)

The rule itself is wrong then. What about fr? Can it be pronounced [vr] (or [vɾ]? I'm not clear on the alternance between trill and tap)? --Per utramque cavernam (talk) 20:29, 18 April 2018 (UTC)

I think fr is always [fɾ] (I can't think of any counterexamples). —Granger (talk · contribs) 20:39, 18 April 2018 (UTC)

susceptible

The -sce- needs to be fixed, I think. Ultimateria (talk) 17:01, 3 April 2017 (UTC)

@Ultimateria You mean the Latin American IPA? Yes, it does. Phonetically, that word has one [s] (but AFAIK not phonemically - /s.s/). Mr KEBAB (talk) 15:51, 6 April 2017 (UTC)

@Erutuon: Hello. I'm bothered with this test case; it's bound to fail, since Module:es-pronunc/testcases returns the Spanish Spanish phonetic transcription, not the Latin American one. It will ultimately return [susθepˈt̪iβle], which still won't be [susepˈt̪iβle].

Is there a way to specify that we're testing the Latin American Spanish output, or to create a second test case page for Latin American Spanish? --Per utramque cavernam (talk) 09:00, 18 April 2018 (UTC)

This is a real problem and needs to be fixed. Can't we just add something to the module that converts "ss" to "s" in Latin American Spanish phonetic transcriptions? —Granger (talk · contribs) 13:05, 18 April 2018 (UTC)

You're describing a different problem (which of course needs to be fixed too): the fact that the current Latin American output is wrong.

The problem I want to tackle is that even if the Latin American module output were right, the testcases page would still be broken, because it outputs Spanish Spanish pronunciations only. To put it another way, you currently can't do testcases for specifically Latin American pronunciations.

In other words there are two things to fix: the module itself, and the testcases page. --Per utramque cavernam (talk) 13:26, 18 April 2018 (UTC)

I could make the module show separate lines for Latin American and Castilian when the two are different, but one otherwise. — Eru·tuon 19:24, 18 April 2018 (UTC)

Doesn't it already do that? By the way, the module has another serious problem – it gives disastrously wrong pronunciations for words like porque and orca. —Granger (talk · contribs) 19:29, 18 April 2018 (UTC)

The template does, but I was talking about the testcases module. Sorry for not being clear. — Eru·tuon 19:32, 18 April 2018 (UTC)

Yes, I think that would work. By the way, it tests the phonetic transcriptions, but not the phonemic one; could there be cases where the phonetic is indeed correct, but the underlying phonemic is wrong, and we wouldn't see it? Are they bijective?

(edit conflict) And martes has the same problem. We are currently displaying incorrect pronunciations in who knows how many Spanish entries. This is a serious problem. —Granger (talk · contribs) 19:33, 18 April 2018 (UTC)

And it (at least the vanishment of the consonant) was easy to fix. But the module now says the trill occurs before the consonants [bdfɡklʎmnɲpstxzʃɟ], whereas the "expected" transcription in the testcases for porque has a tap. — Eru·tuon 19:52, 18 April 2018 (UTC)

Thanks for that. The tap/trill issue does need to be fixed, I think, but it's less serious than the vanishing consonant. —Granger (talk · contribs) 19:59, 18 April 2018 (UTC)

Must fix

word = mw.ustring.gsub(word,"c([ieíé])",(LatinAmerica and 's' or 'θ') .. "%1")

África

It's producing /ˈfɾika/. Also, can someone please address the previous two issues? Ultimateria (talk) 16:43, 29 October 2017 (UTC)

@Ultimateria: This one is an easy fix: you just have to write {{es-IPA|áfrica}}. The module could certainly take care of the substitution, though I don't know how. --Per utramque cavernam (talk) 21:36, 17 April 2018 (UTC)

Btw I've added some testcases for the problems you mentioned; do they seem good? --Per utramque cavernam (talk) 21:37, 17 April 2018 (UTC)

Yes, that's perfect! Bless you for doing this haha Ultimateria (talk) 21:50, 17 April 2018 (UTC)

Problem with K-t clusters

• capturar:
• IPA^(key): /kabtuˈɾaɾ/ [kaβ̞.t̪uˈɾaɾ]
• abyecto:
   

• IPA^(key): /abˈʝeɡto/ [aβ̞ˈʝeɣ̞.t̪o]

• IPA^(key): (everywhere but Argentina and Uruguay) /abˈʝeɡto/ [aβ̞ˈʝeɣ̞.t̪o]
• IPA^(key): (Buenos Aires and environs) /abˈʃeɡto/ [aβ̞ˈʃeɣ̞.t̪o]
• IPA^(key): (elsewhere in Argentina and Uruguay) /abˈʒeɡto/ [aβ̞ˈʒeɣ̞.t̪o]

Both show the voiced/approximate phoneme rather than the voiceless one. That can't be as intended. Coreydragon (talk) 09:42, 29 May 2017 (UTC)

accidental has the same problem. —Granger (talk · contribs) 20:27, 17 April 2018 (UTC)

@Coreydragon, Mx. Granger: I've added some testcases at Module:es-pronunc/testcases; are they correct (I've no idea)? --Per utramque cavernam (talk) 21:18, 17 April 2018 (UTC)

Unless I'm very much mistaken, capturar should have a [p] and accidental should have a [k]. —Granger (talk · contribs) 21:25, 17 April 2018 (UTC)

@Mx. Granger: All right, I wasn't sure about that; fixed. By the way, is it ok if I move your remarks to Module talk:es-pronunc? It's probably better to have everything in a single place. --Per utramque cavernam (talk) 21:42, 17 April 2018 (UTC)

Thanks. susceptible should have a [p] too, and in Latin American Spanish it has only two [s]'s, not three (contrary to what is displayed in the entry). And sure, moving the discussion seems fine to me. —Granger (talk · contribs) 21:49, 17 April 2018 (UTC)

And abyecto should have a [k]. —Granger (talk · contribs) 21:52, 17 April 2018 (UTC)

I'm in complete agreement with Granger, at least in the varieties of Spanish with which I'm familiar (there are, after all, so very many); it's my fault for not having specified the correct pronunciation to begin with. I'm not somewhere where I can whip out a Spanish grammar or phonological sketch at the moment, and it is hard to search for videos specifically where these double-plosive clusters will appear, but in this video you can hear several instances: 2 cases of [kt] right after 0:45, and one of [kθ] around 1:20. I'm sure in any hispanophone video of significant length examples will arise. Thanks Per utramque cavernam —Coreydragon (talk) 21:57, 17 April 2018 (UTC)

Must introduce exceptions to

syllables[i] = mw.ustring.gsub(syllables[i],'[ptk]$',{['p']='b',['t']='d',['k']='ɡ'})

See w:Spanish phonology#Obstruents for a reference for this. I doubt that it makes sense to say that some words follow the rule and some don't. I do see pronunciations that follow this rule on Forvo here (pronunciation by Nakascit) and here, but some do not. Perhaps it's a variable feature. Pinging @Mr KEBAB, who probably knows more. — Eru·tuon 22:59, 17 April 2018 (UTC)

The "Nakascit" pronunciation sounds like an unreleased [k̚] to me. The "hackphet" one does sound like [x] to me, which is surprising, but I notice that it is labeled "Male from Chile", and Chilean Spanish is notoriously different from other varieties. The other pronunciations on the first page you linked to ("cevalenz", "LauraPlatas12", and "carmen_ms") are clearly a stop rather than a fricative or approximant. —Granger (talk · contribs) 23:29, 17 April 2018 (UTC)

I thought I heard [ɣt] in the Nakascit pronunciation, but I don't know if my perception is accurate. It might be easier to determine by looking at the soundfile with software. — Eru·tuon 20:19, 18 April 2018 (UTC)

deshelar

Is it really pronounced [d̪ezeˈlaɾ]? es.wikt has [d̪eseˈlaɾ]. Ultimateria (talk) 17:20, 16 December 2017 (UTC)

@Ultimateria No, of course not. The correct pronunciation is [d̪eseˈlaɾ], with a voiceless [s]. The voiced allophone [z] can only ever appear immediately before voiced consonants and AFAIK only in the syllable coda (so definitely not in deshelar). Only in non-standard accents can [z] appear in other positions. Mr KEBAB (talk) 06:11, 11 March 2018 (UTC)

@Mr KEBAB: Thank you, that's what I suspected. Is this phenomenon predictable enough to put in the code? Ultimateria (talk) 20:36, 15 March 2018 (UTC)

@Ultimateria: I guess so. I don't even know why it shows as voiced. <h> is never [h] in native words, and [h] isn't even a voiced consonant! This is a very strange bug. Mr KEBAB (talk) 02:21, 16 March 2018 (UTC)

@Ultimateria, Mr KEBAB: This one is fixed, though I hope I haven't broken other things in the process. --Per utramque cavernam (talk) 22:05, 17 April 2018 (UTC)

I guess I should even do this. --Per utramque cavernam (talk) 22:21, 17 April 2018 (UTC)

Now I'm getting /z/ in desmentir. Ultimateria (talk) 01:05, 25 April 2018 (UTC)

@Ultimateria: do you mean it should be [d̪esmẽn̪ˈt̪iɾ] instead of [d̪ezmẽn̪ˈt̪iɾ]? Why is it different than mismo? (the testcases page states the pronunciation is [ˈmizmo]). --Per utramque cavernam 12:12, 22 May 2018 (UTC)

@Per utramque cavernam: I believe the prefix "des-" and other prefixes ending in s (trans-, pos-, bis-) maintain /s/ in most situations. Ultimateria (talk) 20:21, 22 May 2018 (UTC)

Discrepancies

• /ˈtʃi.le/ vs. /ˈt͡ʃile/ (diff). (do we want explicit syllabation?)
• [ˈkõ̞ɲ.ɟʝu.xe̞] vs. [ˈkõɲɟ͡ʝuxe] (diff).
• [ẽ̞mˈbjaɾ] vs. [ẽmˈbjaɾ] (diff).

--Per utramque cavernam (talk) 21:05, 17 April 2018 (UTC)

prorrogado

It's producing /pɾoɾɾoˈɡado/, [pɾoɾɾoˈɣaðo]. It should be /pɾoroˈɡado/, [pɾoroˈɣaðo], no? Ultimateria (talk) 03:28, 18 May 2018 (UTC)

Of course. I think some linguists would agree with the phonemic transcription with two taps in a row, but I'm pretty sure it's not the standard analysis. The phonetic transcription is just wrong. This is a serious error and needs to be fixed. —Granger (talk · contribs) 09:44, 19 May 2018 (UTC)

Post-lateral voiced plosives

This generates the plosive allophone of the voiced plosives after /l/, when the fricative/approximant allophone is blocked only for /d/. That is, algo appears /ˈalɡo/ without giving phonetic [ˈalɣo], and volver gives /bolˈbeɾ/ but not [bolˈβeɾ]. For more poignant examples, where the wrong phonetic form is generated, rather than none, hidalgo generates /iˈdalɡo/, [iˈðalɡo] rather than the correct /iˈdalɡo/, [iˈðalɣo], and albergue yields /alˈberɡe/, [alˈberɣe] rather than /alˈberɡe/, [alˈβerɣe]. —Coreydragon (talk) 22:38, 3 June 2018 (UTC)

ununquadio

<qua> is producing /ka/ and not /kwa/. Ultimateria (talk) 12:42, 17 August 2018 (UTC)

This script needs some work

For starters it could stop using </ɟ͡ʝ/> in place of the more normal constrictive /ʝ/ (used by far more authors when describing the phonic structure of Spanish), on this context I will also mention that [aβˈɟ͡ʝeɣt̪o] for "abyecto" is not normal but emphatic pronunciation, though [ˈkõɲɟ͡ʝuxe] is correct.

I disagree with keeping different phonemic transcriptions for different major accents whenever /ʎ/ appears For /ʎ/ this particular phoneme is also not even necessarily linked to the neutral pronunciation of each region, a few accents in America and Spain use it, while it has tended to merge with the palatal constrictive in Spain as well, to the point of not being really a matter of the American-Peninsular divide at this point.

For /θ/, when in contact with voiced consonants, it becomes [ð] (this is of course in Spain, where you also have a larger tendency to voice /s/ as well in contact with voiced consonants)

Now when giving different pronunciations for the two regions, something could be added regarding the apico-alveolar pronunciation of /s/ in Spain, while America tends to have lamino-dental articulations (even if we have tons of variations). The same for /x/, which in America tends to be [x] mostly, a velar approximant, or even [h], while Spain can have [X] mostly, though there can be uvular constrictive trills as well.

Again on /s/, the module shouldn't be giving out transcriptions like /sussebˈtible/, [susseβˈt̪iβle] for America, words like this don't contain geminate /s/ save for people who pay attention to spelling, and even then it's not systematic.

/sr/ sequences have a slight problem: they don't give out stuff like [izɾaˈel], but [iɾr:aˈel] (this also happens for intervocalic /r:/ as well). /s/ in contact with /m, n, ɲ w l, ʎ/ don't necessarily become voiced in both accents, you can have a tendency for voicing in Spain, while America may tend to keep it voiceless. Sequences of /nsC/ like in 'construir' present a problem with /n/, it is not pronounced as [n] in this context normally, but as a semi provelar nasal (which can be roughly translated to offIPA as [ŋ̟] with lowered articulation, so a transcription as [ˈkoŋ̟st̪ɾwiɾ] would be closer to reality than [ˈkonst̪ɾwiɾ] even if not totally on the mark.

Finally, even though only three of the 'diphthongs' in the Template examples are actual diphthongs (there is not diphthong in [ˈdjos] or [weβo], the phonetic transcriptions only work for slow speech, in normal speech the vowels /e, a, o/ in /CeV, CaV, CoV/ contexts will transform into their semi-approximant taxophones (bear in mind these are not vowels either but consonants). Thus you instead get [ˈpw̞eta], even if the phonemic transcription is /poˈeta/, [bj̞atiˈtu(ð̞)] for /beatiˈtud/ and [ɰ̟ ̞oˈɾita] (roughly) for /aoˈɾita/. — This unsigned comment was added by 200.44.66.135 (talk) at 03:30, 1 October 2018 (UTC).

I fixed abyecto. That was just a bug in the module; it wasn't supposed to have the harder allophone of y. — Eru·tuon 04:51, 1 October 2018 (UTC)

/ʎ/ vs. /ʝ/ - Suggestions for fixing this template

I agree that the simpler /ʝ/ (used by far more authors when describing the phonic structure of Spanish) would be a better choice for this template, since it's the phoneme, with huge variations throughout dialects. Also, the two ways of pronouncing ⟨ll⟩, as /ʎ/ vs. /ʝ/ (yeísmo) do not correspond to Castillian vs. Latin American Spanish. There are parts of Spain where /ʝ/ is far more common than /ʎ/. And there are vast parts of the Andean region where /ʎ/ is far more common for ⟨ll⟩. See audio files in llanta for a possible fix (with yeísmo and without yeísmo). Also, you may check https://es.wiktionary.org/wiki/Wikcionario:Referencia/ES/Pronunciaci%C3%B3n#Ye%C3%ADsmo for a map of its distribution. Thanks! --Ahoraes (talk) 22:28, 7 November 2018 (UTC)

explícale

Stress stays on the first syllable when specified as |1=ecsplícale. Ultimateria (talk) 01:54, 23 October 2019 (UTC)

Added to testcases. — Eru·tuon 03:40, 23 October 2019 (UTC)

/kt/, /ks/

The template voices the velar, even though the second consonant is unvoiced, and then the resultant voiced stop becomes the allophonic voiced fricative in the phonetic transcription. I'm not a native speaker, but I'm pretty sure this is wrong. Thoughts? embryomystic (talk) 20:11, 16 February 2020 (UTC)

Coda r

@Erutuon: puerto is rendered as /ˈpwerto/, [ˈpwert̪o]. Shouldn't it be /ˈpweɾto/, [ˈpweɾt̪o] with a flap rather than a trill? —Mahāgaja · talk 12:20, 22 February 2020 (UTC)

@Mahagaja: I think it can be either in that position, and somebody decided to choose the trill over the tap. — Eru·tuon 18:34, 22 February 2020 (UTC)

Yeísmo and no yeísmo

Hi. Could someone change that? It was already requested in 2018. In Template:es-IPA ll is shown as a consonant with different pronounciation in Castile and Latin America but that is not accurate. yeísmo and lleísmo exist in both regions, yeísmo is preferred in both too. Words like llamar should say "(yeísmo) IPA(key): /ʎaˈmaɾ/ (lleísmo) IPA(key): /ɟ͡ʝaˈmaɾ/. Thanks in advance. Lin linao (talk) 18:59, 7 June 2020 (UTC)

Include optional voiceless palato-alveolar fricative for CH/SH or custom IPA

Some dialects—western Norteño, Panamanian, Andalusian and Cuban— would expect a [ʃ] sound instead of a [t͡ʃ]. As example word, the Norteño "chilo"/"shilo". When using {{es-IPA|shilo}} or {{es-IPA|chilo}}, How to remove that initial glottal stop? it should look like: /ʃilo/

Would it be possible to include a dialect selector? E.g.: {{es-IPA|peninsular|chilo}}, {{es-IPA|latam|chilo}}, {{es-IPA|western norteno|chilo}}

Hectormgerardo (talk) 07:23, 11 October 2020 (UTC)

Multiword terms

@Erutuon, PUC Could the module be edited to put spaces into multiword terms like caja de herramientas? The output is just a little hard to read when it's all run together. —Mahāgaja · talk 21:25, 12 January 2021 (UTC)

@Mahagaja: I'm sorry, I don't have the needed coding skills. But @Benwing2 said they'd be willing to have a look at this. PUC – 21:28, 12 January 2021 (UTC)

Fixing this module

(Notifying Ungoliant MMDCCLXIV, Metaknowledge, Ultimateria): @Fay Freak, Embryomystic, Mahagaja, Erutuon, PUC Apologies for the wide pings; I'm not sure who is knowledgeable about Spanish phonology.

I am planning to do some work on this module. However, I have some questions that come from going through the places where manual pronunciations are indicated and fixing them to use the automatic {{es-IPA}} where possible. Hopefully someone is familiar enough to answer. Ideally, we could find a native Spanish speaker; not sure if we have any such regular editors here. Here are my questions:

1. Does x = /ks/ [ks] or /gs/ [ɣs]? I have seen it explicitly written in introductory Spanish books that the latter is normal, and Wikipedia says the following:

   Both in casual and in formal speech, there is no phonemic contrast between voiced and voiceless consonants placed in syllable-final position. The merged phoneme is typically pronounced as a relaxed, voiced fricative or approximant,[53] [quoting Navarro Tomás 1918] although a variety of other realizations are also possible. So the clusters -bt- and -pt- in the words obtener and optimista are pronounced exactly the same way: obtener /obteˈner/ > [oβteˈner]; optimista /obtiˈmista/ > [oβtiˈmista].

   However, most of the manual phonetic transcriptions use [ks], not [ɣs]. Which is correct?
2. Same with ct, = /kt/ [kt] or /gt/ [ɣt]? Again, most manual phonetic transcriptions indicate [kt].
3. Same with cc before front vowel in distinción Spain (e.g. conducción), = [kθ] or [ɣθ]?
4. Same with cn as in técnica: [kn] or [ɣn]?
5. On a similar note, what about final -t, -c, as in acimut, Chapultepec? Wikipedia would imply this should be final /d/ [ð] and final /g/ [ɣ], and that's what the module does; but many phonetic transcriptions give [t] and [k], and in fact many uses of {{es-IPA}} respell words with final -t using -th to force [t].
6. On a similar note, does absceder have [ab] or [ap] (the latter indicated in manual transcription)? What about adoptada, with [pt] or [βt]? What about algoritmo, with [tm] or [ðm]?
7. On a similar note, what about tl in atlántico? [tl] or [ðl]? This is potentially complicated by borrowed Nahuatl terms like chipotle. Do these behave the same as atlántico? What about syllable-initial tl- as in tlacuache (given as /tɬa.ˈkwa.tʃe/) tepezcuintle etc.? Is [tɬ] correct?
8. What about tz in borrowed Nahuatl terms like quetzal, and potentially also in borrowed Basque terms? [ts] or [ðs]? Also cf. tzompantli given as /somˈpan.tɬi/|/tsomˈpan.tɬi/.
9. sc + front vowel, e.g. in descendiente: The manual transcription of most words indicates that this is [sθ] in Spain (distinción), [s] in Latin America; but lascivia gives [(s)θ] in Spain, [(s)s] in Latin America. Which is correct?
10. Does Addis Abeba have double /dd/? What about cappa/kappa, does it have double /pp/?
11. Some pages include accents other than standard Spanish distinción and standard Latin American. What do we do about this? The most common other accent indicated is Rioplatense. Manual transcriptions randomly indicate medial -y- and -ll- as [ʒ], [ʃ] or both, and tend to indicate coda -s as [h], which is far from unique to Rioplatense. When I was in Buenos Aires, I only heard [ʃ], but I have heard that [ʒ] is the older pronunciation in Buenos Aires and is maybe still the norm in other places, e.g. Uruguay. Do we bother including these pronunciations?
12. In a similar vein, there are some pages that include transcriptions of -j- as [h], marked as "Caribbean/Venezuelan" or simply "some areas of Latin America". Examples are carajito, jojoto, Juan, jeva, mujer. Do we include these?
13. mes has the pronunciation [mɛː] listed, tagged as "Murcia, Eastern Andalusia"; millo "maize, corn" (a Canary Islands term) has /ˈmiʝo/ given, tagged as "Canary Islands" although this pronunciation is actually the norm almost everywhere; chilo has /ʃilo/ given, tagged as "Norteño Western Spanish". Do we include these?
14. Initial ps-, pt-: Most manual transcriptions indicate that p- is silent, but some include it or give it as optional. What is correct?
15. French terms: baguette = /baget.te/ or /baget/ or /baged/? ballet = /balé/, /balet/ or /baled/?
16. biodegradable = /bi.odegradable/ or /bjodegradable/ or both?
17. Any ideas on caballo de Przewalski? The transcription had /sebalski/ but I assume in Spain it would be /θebalski/ (maybe?).
18. desnudo: two pronunciations given [de̞sˈnuðo̞] [de̞zˈnuðo̞]. Correct?
19. Hiatus in león, michoacano, no hablo inglés, sexo oral, toro embolado, torre Eiffel, vehemencia, vía crucis, vivito y coleando, Vuelta a España, zoo. Manual transcriptions given as /leˈon/|/leo̯n/, [mi.t͡ʃoa̯ˈka.no]|[mi.t͡ʃwaˈka.no], [no̯‿a.βlo̯‿iŋˈɡles], [ˌsek.soˈɾal], /ˈto.ɾo em.boˈla.do/|[ˌto.ɾoe̯m.boˈla.ðo], /ˌto.re‿ei̯ˈfel/, /be̞ːˈme̞nθja/ (distinción), /ˈbi.a ˈkɾu.θis/|/bjaˈkɾu.θis/ (distinción), /biˈβi.toi̯ koˈlea̯n.do/, /ˌbwel.t(a) a esˈpa.ɲa/, /ˈθo.(o)/ (distinción). (a) Are these correct, and (b) if so, do we want to indicate this sort of hiatus reduction?
20. mongol given as [moŋ.ˈɰol], tambor given as [ta̠mˈβ̞o̞ɾ]|[ta̠mˈbo̞ɾ], velde given as [ˈbe̞lðe̞]. Seems wrong to me, I thought voiced stops are always hard stops after nasals, also d is hard after l. Comments?
21. exacto, excelente, excusa given with optional /k/: (1) /e(k).saɡ.to/|[e(k).saɣ̊.to]; (2) Latin America /ekseˈlente/, Spain /e(k)sθeˈlente/; (3) [ekˈskusa]|[esˈkusa]. Correct?

Sorry to give such a large list. Feel free to refer to the above entries by number when referencing them. Benwing2 (talk) 03:49, 25 February 2021 (UTC)

Why haven't you pinged any natives? @Koszmonaut, Pablussky, AugPi, Ser be etre shi, Vivaelcelta are all active. I could help with this, but I'd rather trust the experts over my own judgement. —Μετάknowledge^{discuss/deeds} 05:27, 25 February 2021 (UTC)

@Benwing2 Hi. So many legitimate questions; I think people are often surprised to read that what they thought was [ks] (or [gz]) is actually [ɣs]. :) Let me start by saying I am in awe by your attention to these issues, and appreciate your interest. Note: I have read some monographs on Spanish phonetics, but not at all a lot. Perhaps it amounts to merely five books plus some articles.

#1-4 and #6: I would say that most speakers in casual speech do naturally say [ɣs βt ɣt ɣn ðm] and perhaps [ɣθ], including [βt] for bt/pt and [βs] in absoluto and opción (in Lat. Am.). However I would say that in conscious, careful speech (say, when consciously pronouncing a word in isolation) many speakers do make an effort to "sound out the letters" and pronounce pt as [pt], cc as [ks] (acción), and so on. Manual phonetic transcriptions often use "ks" instead of "ɣs" likely as an influence from spelling or as a "broad" transcription closer to phonemes.

#5: wow, those should have [t] and [k], yes. [#6 covered with #1-4.] #7: this has long been a source of mystery for me. I was so shocked to read about that in a couple books, that I've been asking some people from Spain about it, and so far none have turned out to have [að.la-] in atlántico. They had [a.tla], as in Mexico. And when I provided them [að.lan-], syllabified as ad-lan-, they have rejected it. Maybe it was true it was [að.la-] when Navarro Tomás published his famous work in the mid-20th century, but I have doubts about the reality of this now... I would recommend using [tl] as an onset. I'd love to hear from Pablussky on this or anyone from out of Mesoamerica.

#8: Interesting; I am not sure what I say myself. Perhaps I simply vary between [ðs] and [ts], as pointed above. It is hard to analyze one's own phonetic speech, as you may know. I haven't seen this cluster mentioned in phonetic works. #9: Phonetic works generally describe this as you say, [sθ] or carefully/artificially [ksθ] in Spain north of Andalusia, and [s] in Latin America. #10: these words have a single /d p/. In truth I have never heard Addis Abeba pronounced, but I have heard kappa, and have no doubts the former would have a single /d/, unless the speaker is imitating Amharic for fun.

#11: I would say Rioplatense is such a notable (and notorious) accent it is desirable to include it. They do have [ʃ]~[ʒ] as the standard surface form of (pan-dialectal) /ʝ/. They also seem to consistently use /s/ [h] with some rules (which, I do not know) in their careful speech, avoiding the pronunciation full of /s/ [s] used in Mexico and sometimes formally in El Salvador (where we have plenty of /s/ [h] in the local dialect).

#12: I vote for leaving them to manual inclusion. In particular, mujer [muˈheɾ] is so predictable from the phonemic transcription "/muˈxeɾ/" that I don't see great usefulness in having it. And I say this as a speaker of a dialect with /x/ [h], including [muˈheɾ].

#13: I vote against including them, due to being too non-standard. Do we have a practise of including non-standard accents in pronunciation templates? Another problem I see is that, in such dialects, words more typical of the written language tend to be pronounced in a more standardized way. Those who say mes as [mɛː] might say obtendrás as [oβtenˈdɾah] or even [optenˈdɾas], as opposed to [-dɾɑː]. I see no way to encode that a word is typically said with a fancy pronunciation or not, which is why I favour automatizing prestigious pronunciations only.

#14: In my experience, a small minority speakers like to say them as a spelling pronunciation, but the norm, by far, is to leave p- as a silent letter. Incidentally, this is true of English consonant phonemes too, e.g. those who say el sushi consistently as [ˈsuʃi] instead of [ˈsutʃi], Javascript as [ˈɟʝaβaskɾipt] instead of e.g. [ɟʝaβaˈskɾip jaβahˈkɾit ˈɟʝaβahkɾih] (English word-final clusters are a true source of variation today, depending on the level of English education, and even attitude towards the English-speaking world).

#15: They generally have [t], even when the -t is silent in French. Certainly ballet does. Same goes for Champagnat [tʃampaˈɲat], as in the last name of the saint whose organization runs Catholic schools in some Lat. Am. countries. I have never been involved in the ballet world, but it wouldn't surprise me people nowadays are pronouncing it [baˈle] under better awareness of French, though.

#16: This is the source of interesting dialectal variation. We say [bjo-] in El Salvador, and I think people do in Mexico too, but I have noticed Rioplatense and Peninsular speakers generally say [bi.o-]. I am not sure whether this phenomenon documented well in phonetic publications.

#17: No clue. I'd fathom most people when facing it would read it [peɾseˈβalski] unless given an "educated", "in-the-know" [seˈbalski], cf. Sigmund Freud, whose last name is indeed generally [fɾoið] among those who have heard of him in a high social context like universities (as opposed to stumbling on him on Wikipedia, say! in which case people might say [fɾeu̯ð]...).

#18: I would say it is correct. Phonetic works always describe /sn/ as [zn], but I find the voicing is so light that I prefer to transcribe it [sn], i.e. this might be down to preference or narrowness of transcription. You may want to stick to [zn] as linguists' publications prefer that. In contrast, I would say /sb sd sʝ sg/ clusters (resbalar, desdicha, las llantas, riesgo) do clearly have [z].

#19: All phonetic publications agree there is some variation in such words. The RAE even admits in its 2010 Ortografía that there is a certain distance between speech and the stringent rules proposed for syllabification, notably in the case of ui (as the RAE itself points out), which the RAE always syllabifies as a diphthong no matter the actual pronunciation, e.g. huir which is generally [u.ˈiɾ] is treated as a monosyllable. León can be [leˈon le̯on] or even colloquially [ljon]. I would say that across words (toro‿embolado, torre‿Eiffel) it is very common and probably desirable to include it, but I'm not sure if I'm being misled by my dialect, which according to Lipski in Latin American Spanish is notorious for its diphthongs across words... Also, I am very surprised by zoo [θo], but I am not from Spain... It's always zoológico [so.oˈlo-] in El Salvador.

#20: You are right; they're wrong. #21: They are correct, cf. #1. Dropping [k] in coda [ks] (excusa, sexto, extrañar, experiencia...) is quite common, but also widely considered "incorrect." I am of two minds regarding whether it should be included. Perhaps it could be indicated as "colloquially, often". Others may have better suggestions, and I wouldn't oppose omitting this, i.e. only providing [eksˈkusa, ˈseksto, ekstɾaˈɲaɾ].--Ser be être 是^talk/stalk 07:30, 25 February 2021 (UTC)

@Benwing2 @Ser be etre shi Warm hello to both of you. I'll add my two cents but I'll also admit beforehand that I'm no expert in phonetics or phonology. My fellow Spanish-speaker above seems much better versed in literature on the subject; that said:

1-4 and 6:: I don’t think I can add anything more. I’m in awe of the exhaustiveness of Ser be etre shi’s answers.

5: Definitely word-final [t] and [k].

7: I’ve never heard [ð.l] in Mexican Spanish. I’ve heard both [t͡l] and [t͡ɬ] for tl, and would and say that those who pronounce [tɬaˈkwatʃe] would also pronounce [aˈtɬantiko], same for [t͡l].

8: I created this particular entry and included the two pronunciations I’ve heard in Mexican Spanish. I’d say most speakers tend to say [s] but that [t͡s] is also a valid pronunciation. For example, zopilote (from Nahuatl, tzopīlōtl) is the standard Spanish spelling and indicates an absence of the word-initial cluster.

9: I would advise against the geminated [s] for Latin American Spanish. I always say [desenˈdjente].

10: I’d say all of these have a single, non-geminate consonant sound.

15: Again, I’m unfamiliar with word-final [d] or [ð] in this context. Def [baˈlet], and I’ve overwhelmingly heard [baˈɣet].

16: I agree with what’s been said. I’d say Mexican Spanish at least tends toward [bjo], but I’m not totally a stranger to pronunciations with the hiatus between the two vowels.

17: I also created this entry and approximated the pronunciation from what I heard in videos concerning the horse, where aapproximations of the Polish pronunciation are used. I agree that most Spanish-speakers unfamiliar with the animal or coming upon a sign at a zoo would have no idea where to begin to pronounce, perhaps [pɾeˈβals.ki], [pɾeˈwals.ki] or even [peɾseˈwals.ki].

18: I definitely always voice /s/ before /n/ and /m/ but agree that [z.n] is appropriate in a closed transcription and /sn/ in an open one.

19: I’ve had run-ins with editors on Wikipedia over this. In a closed transcription I’d say [mit͡ʃoa̯ˈkano] and the like have a diphthong made up of strong vowels. San Miguel de Allende, to give another example, is [dea̯] in Mexico, but Wikipedia currently has [de a].

20: Agree with what's been said.

21: Again, agree with the comments above. I’d just point out that it's [s.k]. Syllable-initial [sk], at least as far as I know, is foreign to Spanish.

—K  ^(talk / _contribs) 22:35, 25 February 2021 (UTC)

@Ser be etre shi, Koszmonaut Thanks to both of you! This is very helpful. I will get to work on the module. BTW I'm thinking of adding syllable markers (.) between each syllable, as is currently done with Latin; see for example administrātor. What do you think? Benwing2 (talk) 03:15, 26 February 2021 (UTC)

Also, the module currently marks distinción and lleísmo pronunciations as "Castilian" and seseo/yeísmo pronunciations as "Latin American". We all know this isn't very accurate. I'm thinking we should instead do one of two things: (1) notate them using the standard Spanish terminology distinción/seseo and lleísmo/yeísmo, or (2) notate distinción vs. seseo as something like "northern/central Spain" vs. "southern Spain, Latin America" and lleísmo vs. yeísmo as "rural northern Spain, Andean" vs. "elsewhere". Both proposals mean that there may be up to three pronunciations indicated, e.g. in words like amorcillo and belleza, but I don't see this as an issue. Option (1) has the advantage of being linguistically completely accurate, but has the disadvantage of being opaque to people not already familiar with Spanish dialects. Option (2) has the advantage of making it clear where (approximately) the different pronunciations occur and avoiding technical terminology, but has the disadvantage of glossing over some of the dialectal complexities. On the whole I prefer (2). I'm also thinking based on the responses to #11 above that we should include a rendering of Rioplatense speech, particularly where y or ll occur. This will make belleza have four pronunciations, but I don't see this as a problem. I'm also thinking instead of "Rioplatense" we should say "Argentina, Uruguay" even though this is a bit inaccurate in that per the map in Rioplatense Spanish, interior northern Argentina is excluded; "Rioplatense" may be opaque to many people. Thoughts? Benwing2 (talk) 03:38, 26 February 2021 (UTC)

Also, I really don't like the current use of phonemic /ɟ͡ʝ/; it's both inaccurate and overly complex (even most people familiar with IPA won't recognize these symbols). Unless there are objections I will change this to /ʝ/. Benwing2 (talk) 03:41, 26 February 2021 (UTC)

@Benwing2 Phonemic transcriptions: I used to be a fan of that for well-known reasons (continúo continuó /kon.tiˈnu.o kon.tiˈnuo/ [kon.tiˈnu.o kon.tiˈnwo]), but then I got many rebuttals for it from people online, for good reasons. First, there's no real reason to try to come up with analyses without morphemes carrying phonemic information and without morpheme boundaries, as interesting an exercise as that can be. Second, much of the advantage comes from the IPA's poor choice of only notating stress before a syllable, instead of before a vowel nucleus too: the problem disappears if you use /kontinúo kontinuó/.

Multiple accents: I agree with every one of your preferences. I value articles being understandable by the non-specialist more than total accuracy. Two issues though. First, distinción is widespread in southern Spain today: nearly every time I hear a young YouTuber from Andalusia, they have /s θ/ quite fine, at most with just a few non-standard uses (e.g. the demonstrative ese as [eθe]), all while having plenty of /s/ [h], and /st/ [tʰ], and padre[ˈpaɾe] vs. pared [paˈɾe], and what not. Surprisingly, it might be more accurate to label distinción as "Spain"—I don't think this is being discussed in publications as of yet though. Second, it's hard to avoid lleísmo, because it exists in both Spain and Latin America, and only in very minor (i.e. non-prestigious) accents too, often influenced by the likes of nearby Basque, Catalan, Quechua or Mapuche (an uncle of mine, from rural Chile, has lleísmo...). I also vote for labelling Rioplatense as "Argentina, Uruguay". The accent is essentially standard in those countries at any rate.

Use of "/ɟ͡ʝ/": Personally I find the choice of /ɟ͡ʝ/ perfectly sensical and accurate, and a valid alternative to "ʝ". It has the advantage of emphasizing its phonetic alignment with /b d g/ in most dialects, whereas /ʝ/ has the advantage of... I don't know, looking more like "y"? Nevertheless, I agree it should be changed to "ʝ", because using "ɟ͡ʝ" this way is very unconventional (I have never seen it in publications), as much as I love it.

@Koszmonaut I think you meant to write [-tʃo̯a- -de̯a-] there though, with the non-syllabic diacritic under the first vowel.--Ser be être 是^talk/stalk 18:55, 27 February 2021 (UTC)

@Ser be etre shi When you say "phonemic transcriptions" above are you referring to whether to add . for syllable boundaries? I just made a bunch of fixes to the module including notating all syllable boundaries with "."; check out e.g. Module:es-pronunc/testcases or Template:es-IPA to see the results. If you don't like it, I can take it out with a one-line change. (Underlyingly, the module computes syllable boundaries in any case in order to correctly place the stress marks; it's just a matter of whether to surface that information.) Benwing2 (talk) 19:17, 27 February 2021 (UTC)

@Ser be etre shi BTW I don't have a problem labeling distinción as just "Spain". I was in Seville about 20 years ago and I don't remember hearing a lot of distinción there, but I have read in Harris and Vincent The Romance Languages (from 1990) that distinción is spreading south, so it doesn't surprise me that speakers in Andalusia are now using it. As for lleísmo, I'm about to implement the multi-way split for distinción/lleísmo + Rioplatense, and we can see how it looks. Benwing2 (talk) 19:22, 27 February 2021 (UTC)

@Benwing2 Yes, I meant the use of dots with phonemes inside slashes (not square brackets). There appears to be a certain convention among linguists to avoid marking syllable boundaries in phonemic transcriptions in slashes (which is why people would react so negatively so often when I did it in Spanish...).

@Koszmonaut Also, I am amused by this problem we have in Spanish of people thinking the RAE's stringent syllabification rules reflect pronunciation. It's likely because almost no one actually reads the RAE's works (lol). Both the Diccionario panhispánico de dudas and the 2010 Ortografía actually clarify there's a difference between orthographic syllabification (used in the assignment of the acute accent and cutting words at the end of a line), which is stringent and largely unambiguous, and how people actually pronounce words. One example is truhán, which is monosyllabic in the orthography, but phonetically usually [tɾuˈan] (disyllabic) though sometimes [tɾwan] (monosyllabic) depending on the dialect. Show them this link next time ("en el habla, la secuencia de dos vocales abiertas —especialmente cuando ninguna de ellas es tónica— puede articularse como diptongo").--Ser be être 是^talk/stalk 19:24, 27 February 2021 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘@Koszmonaut, Ser be etre shi A few more questions:

1. obvio: one or two /b/'s or both are possibilities?
2. Cuauhtémoc, buey, veinte, deuda: currently we indicate a glide before a vowel as /j/ /w/, but a glide after a vowel as /i/ /u/, cf. Cuauhtémoc /kwauˈtemok/, [kwau̯.ˈt̪e.mok]. Do we want to do this? Is there any phonetic distinction involved?
3. muy: Currently given as phonetic [ˈmui̯], this seems wrong. Update: I fixed this.
4. Stress in single-syllable words (e.g. muy just above): do we want to indicate this?
5. Stress across multiple words: Currently the module attempts to convert stresses on non-final words to secondary stress, hence ojo rojo -> /ˌoxo ˈroxo/, [ˌo.xo ˈro.xo]. It has a lot of issues currently, cf. la dueña -> /la ˈdweɲa/, [la ˈðwe.ɲa] but el dueño -> /ˌel ˈdweɲo/, [ˌel̪ ˈd̪we.ɲo], dar pie -> /ˈdaɾ pje/, [ˈd̪aɾ pje] but dar plantón -> /ˌdaɾ planˈton/, [ˌd̪aɾ plãn̪.ˈt̪õn]. What is the correct behavior? I'm thinking maybe we indicate secondary stress on each word but the last one, which takes primary stress; but we maintain a list of words that are inherently unstressed: el, la, los, las; un; me, te, se, lo, le, nos, os; mi, mis, tu, tus, su, sus; que, si; y, e, o, u; de, del, a, al; por, en, con; maybe I missed some. Spanish makes this easier by automatically putting stresses on words like él, sí, tú that would otherwise be homonyms of these words. Occasional cases like la del pulpo would just require a stress mark to be placed on a stressed word like la in the respelling. Another possibility is to do as above but dispense with the secondary stress marks and just mark every stressed syllable with a primary stress. Update: I implemented this, with secondary stress marks except for the last primary stress.
6. agriamente, rápidamente: Currently the module generates /aɡɾjaˈmente/, [a.ɣɾja.ˈmẽn̪.t̪e] but /ˌrapidaˈmente/, [ˌra.pi.ða.ˈmẽn̪.t̪e]. I assume the latter is correct and not the former, which needs a secondary stress marker? Update: I fixed this.
7. I'm thinking we should use a lowering mark on voiced fricatives in the phonetic variant to indicate that they are actually approximants: β̝ ð̝ ɣ̝. Thoughts? Update: I implemented this.
8. Double letters: (1) atto- is respelled aththo, which strongly suggests a double t. Correct? (2) perennifolio is given the pronunciation /pe.ɾe(n).ni.ˈfo.ljo/, with optional double n. Correct? What about inmigración? Currently the module generates two double m.
9. software: The page had manual /ˈsof.weɾ/; I changed it to use the module with respelling either sóftwer or sófwer. Correct?
10. istmo: The page had manual /ˈis(t).mo/; I changed it to use the module with respelling either istmo or ismo. The latter generates phonetic [iz.mo]. Correct? Benwing2 (talk) 06:33, 28 February 2021 (UTC)

Benwing2 (talk) 21:36, 27 February 2021 (UTC)

#1: Ah, spelling pronunciations... I'd say people commonly say [ˈo.βjo], but [ˈob.βjo] is also a pronunciation you hear, due to the spelling. Natives who are able to articulate [v] may also use [ˈob.vjo], especially emphatically...

#2: That simply follows the common practice among phoneticians. For reasons I personally fail to understand, many are uncomfortable with notating rising diphthongs as e.g. [ˈbu̯eno]—something to do with coarticulations in onsets... Conversely, falling diphthongs are not that commonly notated with [j w], especially [j], e.g. [ˈbajle]. I would say this is desirable, for the sake of conventions, but we could also just have [i̯ u̯]. After all, colloquial Spanish doesn't really distinguish between /gw w u̯/, e.g. hueso can be [ˈgweso] (and [ɣ̞weso] with lenition: trae huesos), although speakers commonly try to follow spelling pronunciations, so, [ˈweso] (but also commonly fail at it).

On that note, I wonder if it'd desirable to indicate words with /b d ʝ g/ can have lenition word-initialy, e.g. yema [ˈɟʝema ʝ-], as in la yema [laˈʝema]. Probably not, as that depends on the previous word, and it's something readers should know about sound alternations.

#3: [ˈmui̯] is correct. The reason even some phoneticians don't like /ui̯/ as a diphthong is that muy is the only common word with it (cuy is uncommon, and uy is an interjection). But I've read at least one work that criticized this, pointing out that when said with emphasis, fui 'I was' is [fwiii] but muy is [muuuui̯]. Fui and muy do not rhyme for me.

#4: We don't. That'd be strange. Monosyllables can be stressed, but only within sentences...

#5: That strikes me as rather questionable and unconventional... although in some way fine. I would probably prefer dispensing of secondary stress marks.

#6: I would say these actually have two primary stresses, [ˈaɣɾja ˈmente], [ˈrapið̞a ˈmente], but there's a fair bit of disagreement in works on Spanish phonetics. I don't know why there is disagreement (but I suspect some linguists are just too influenced by spelling...). As far as I know, we write them as one word because of French and Italian. Grammatically, they even behave separately, as when coordinating two -mente adverbs: agria y rápidamente, feroz pero dolorosamente. It's not an artificial thing. We naturally say these as [feˈɾosˈmente], [doloˈɾosaˈmente], [feˈɾos i doloˈɾosa ˈmente], etc.

#7: Good!

#8: I would say no one pronounces double tt as [tt], but nn is indeed very commonly pronounced [nn], even in roots e.g. perenne [peˈɾen.ne]. /nn/ is, strangely, the only phonemic geminate consonant in Spanish. Inmigración is pronounced [in.mi.ɣɾaˈsjon]. (I am surprised by how unhelpful Spanish Wiktionary is on these matters... Shouldn't it have pronunciations for these things? Maybe there's just way less activity...)

#9: English loanwords are a source of great variation. Compounds are very commonly pronounced with stress on the last component, especially by those less adept at English (the majority), e.g. [sofˈweɾ]. But you also do get [ˈsof.weɾ], [ˈsoft.weɾ].

#10: Yes, correct. Both [ˈis.mo] (i.e. [ˈiz.mo]) and the closer-to-spelling [ˈist.mo] are in use and considered essentially correct.--Ser be être 是^talk/stalk 06:54, 1 March 2021 (UTC)

@Ser be etre shi I have implemented support for multiple variants, as you can see e.g. in cebolla (which has the maximum of 6 variants listed). Thanks for your detailed comments above, I will go through them and clean things up accordingly in the next day or so. Benwing2 (talk) 06:09, 8 March 2021 (UTC)

@Benwing2 I appreciate your work on this, but isn't listing 6 varieties potentially confusing? It's also taking up a lot of space in the entry, especially when there's only a single definition line. Maybe the rarer pronunications could be collapsed, similar to the pronunciation section in Ancient Greek entries? – Jberkel 09:08, 8 March 2021 (UTC)

Agreed. Even just indenting them under Spain and Latin America would look much neater. Ultimateria (talk) 17:20, 8 March 2021 (UTC)

And in the same vein the accent labels at yo are really wordy. Could the first one go label-less with the other two indented underneath? Ultimateria (talk) 17:43, 8 March 2021 (UTC)

Also in favor of a collapsable pronunciation section as those used in Ancient Greek entries. Also, @Ser be etre shi, wondering why {{es-IPA}} generates a syllable marker before the stress marker. Is this necessary? —K  ^(talk / _contribs) 23:23, 8 March 2021 (UTC)

@Jberkel, Ultimateria, Koszmonaut, Ser be etre shi OK let me see about hiding the less common varieties by default. As for the syllable marker before the stress marker, I thought about whether to suppress that, not sure so I left it for now. Benwing2 (talk) 03:11, 9 March 2021 (UTC)

@Jberkel, Koszmonaut, Benwing2 I would say it would be best to remove the syllable separator before the stress mark, yes. It's just a conventional IPA thing to do so.--Ser be être 是^talk/stalk 06:26, 9 March 2021 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ @Jberkel, Ultimateria, Koszmonaut, Ser be etre shi I have fixed the module to implement collapsed pronunciations, so that only "Spain" and "Latin America" appear by default. The collapsing is only present when there is something to collapse, so you will find it on a word like cebolla, llover, yo or hielo but not suceder or ahora. I had to remove the bullet from before {{es-IPA}} and have it generated by the template itself. I've changed all the Spanish lemmas to remove the bullet, but the non-lemma forms are still to be done. Benwing2 (talk) 07:01, 14 March 2021 (UTC)

@Ser be etre shi Looking really good! Thanks. Quick slightly off-topic question: Do you work on pronunciation templates only? I need some help sorting out entry templates for Old Spanish, if that's possible. (I'm guessing there's another page for discussing the matter further if possible; not my intention to derail the conversation at hand). —K  ^(talk / _contribs) 04:32, 15 March 2021 (UTC)

@Benwing2: Thanks, all those newer changes are very welcome! I trusted you'd remove the bullet soon, though no one pointed it out here. @Koszmonaut: I'm not the programmer who's been editing this module; Benwing2 is.--Ser be être 是^talk/stalk 01:23, 18 March 2021 (UTC)

@Koszmonaut By entry templates do you mean headword templates? I only see one, which is Template:osp-noun. We can discuss on the talk page there if you want. Benwing2 (talk) 01:30, 18 March 2021 (UTC)

Weird behavior

@Benwing2 On entries with the phonemes /j/ or /ʎ/, e.g. royal#Spanish, calle#Spanish, yanqui#Spanish, saya#Spanish, etc., whatever in the module that allows the display to "open" and "close" is also stretching the entry out horizontally to a ridiculous extent. —Mahāgaja · talk 21:02, 20 March 2021 (UTC)

@Mahagaja By "stretching the entry out" do you mean it is right-justifying? I will see if I can fix that. Benwing2 (talk) 21:44, 20 March 2021 (UTC)

@Benwing2: I can't tell if it's right-justifying, but the category names, for example, extend way to the right off the normal screen, and if I want to open the pronunciation "flap", I have to scroll vertically quite some way to get there. —Mahāgaja · talk 21:47, 20 March 2021 (UTC)

@Mahagaja Hmm. For me, it right-justifies the word "more" and everything inside is left-justified. What browser are you on? Are you on a mobile device? Does it happen the same on a different browser? Benwing2 (talk) 21:49, 20 March 2021 (UTC)

@Benwing2: Yes, the word "more" is all the way to the right, but you have to horizontal-scroll to even find it. It happens across browsers on my desktop, but only when Tabbed Languages is activated. —Mahāgaja · talk 21:59, 20 March 2021 (UTC)

@Mahagaja: OK, I will see if I can remove the right justification. I'm not familiar with the Tabbed Languages gadget, which must be why I don't see it. Benwing2 (talk) 22:24, 20 March 2021 (UTC)

@Benwing2: Maybe take a look at how Module:grc-pronunciation deals with the placement of the "More" flap; that one works fine for me. —Mahāgaja · talk 09:56, 21 March 2021 (UTC)

@Mahagaja Please take a look now. I did copy the handling of Module:grc-pronunciation but it looks like I got the length computations wrong. Please review a few different entries to make sure the widths are all correct. Benwing2 (talk) 20:20, 21 March 2021 (UTC)

@Benwing2: I checked several, and they all look OK to me. Thank you! —Mahāgaja · talk 20:23, 21 March 2021 (UTC)

words ending in -y

@Benwing2 An IP had to add manual stress to e.g. maguey and chamoy because they weren't stressed on the final syllable for some reason. Ultimateria (talk) 17:40, 19 August 2021 (UTC)

Stress on words ending in y

The module is currently giving the wrong stress for multisyllabic words ending in -ay, -ey, -oy, -uy (such as magüey, estoy, Jujuy); these should have ultimate stress but they are coming out as penultimate. I have no idea how the code works but I suspect that what's happening is that something is changing -Vy to -Vi – presumably to generate a diphthong instead of a /ɟ/ – and then assigning stress based on the changed form with -Vi (which ends in a vowel and is thus penultimately stressed) rather than the original -Vy.

I've manually corrected a couple but this should be fixed at module level. 5.81.101.117 17:52, 19 August 2021 (UTC)

• Corrected. Burzuchius (talk) 19:44, 15 October 2021 (UTC)

“Lleísmo”

This is misnomer. The term “lleísmo” does not properly refer to the y–ll distinction (which is the original state of affairs, just like with the c/z–s distinction, and therefore merits no special name or designation, as it does not constitute any kind of phonological innovation), but instead it refers to the confusion of the original <y> and <ll> phonemes into a merged phoneme with the sound of original <ll> (that is, basically just like “yeísmo” but with the opposite end result). Admittedly, this true, properly-called “lleísmo” is an exceedingly rare phenomenon in Spanish (but it has been nonetheless documented, in some rural parts of, I think it was somewhere in the mountainous areas of the Cáceres province, if memory serves).

OTOH, the template also generates an inadequate label for this y–ll distinction misnamed “lleísmo” (which should be properly called “distinción” instead, that is: “yeísmo” vs “zheísmo” vs “sheísmo” vs “lleísmo” vs “distinción”, just like with “seseo” vs “ceceo” vs “distinción”). The generated label qualifies this pronunciation as a “rural northern Spain” phenomenon, which is a VERY inaccurate description of the geographic extent of the y–ll distinction.

For starters, some of the most prominent strongholds where this phonological distinction remains nowadays are not located anywhere in Spain anymore, but in South America (particularly in Paraguay, the rural Andes, and the Argentinian areas in between those) and in Philippines Spanish, where this phonological distinction has always been very alive and well (even though—paradoxically—the Philippinesʼ connection with Spain took place through the port of Acapulco in Mexico, where Spanish dialects are “yeísta” since long ago). In fact, the y–ll distinction has been so strongly maintained in the Spanish of the Philippines, that it was kept in Spanish loanwords borrowed into native Philippine languages (such as Tagalog “kalye” from Spanish “calle”, or Cebuano “lyamar” from Spanish “llamar”), and it exists also in the Spanish-based creole Chavacano.

Plus, within Spain itself, the original y–ll distinction was never limited to the rural northern areas (for example, there used to be y–ll distinction until modern times documented in some parts of Spain as far south as Andalusia, such as some localities in the province of Seville). Even though, admittedly, the modern “yeísmo” (as opposed to the much older “etymological yeísmo”, the one that as its end result produced the modern <g/j> in words such as <ojo> from oculum > ocʼlu > ollo > ozho > osho > ojo, or <mujer> from mulierem > muliere > muller > muzher > musher > mujer, and which uniformly permeated all dialects of Old Spanish since its end results are visible everywhere in the modern varieties including Judeo-Spanish) certainly was historically a more southern-than-northern thing (modern “yeísmo” started to appear in documents from the old Kingdom of Toledo at the end of the 14th century, and quite soon expanded mostly southward towards nowadays Andalusia and a bit northward towards nowadays Madrid, quickly claiming large parts of the southern half of Spain. Whereas the Old Castile (north of the Sistema Central), as well as the peripheral bilingual areas, took much longer to get infiltrated by this modern “yeísmo”, with most of these not becoming strongly and majoritarily “yeísta” until the mid-to-last 20th century (when, between the influence of the radio and television broadcast from traditionally “yeísta” Madrid, and the large population migrations from poorer Andalusia and Extremadura towards the capital and towards the wealthy industrial poles in the north during the 1950s–60s–70s, the speech of most urban areas throughout the country was quickly turned from clear distinction to almost total “yeísmo” in the matter of just a few decades, plus with the migration of most of the younger rural populations everywhere towards the now-strongly-“yeísta” cities, the y–ll distinction still kept alive among the older rural populations in historically distinguishing areas has finally become an obsolescent feature almost throughout Spain and unfortunately will expectably all but disappear completely here during the present 21st century, just like it earlier did in French or in Hungarian, where their equivalent “yeísmo”-like phenomena are now the only recognized standard pronunciation, with only a few remnant relic populations of speakers retaining the original-but-no-longer-standard pronunciations of French <-ill(e)> and Hungarian <ly>). 92.178.80.48 18:24, 29 March 2022 (UTC)

In response to this IP, the term lleísmo referring to the maintenance of an ll-y distinction is relatively common; see Wikipedia's Yeísmo article (with three citations of this usage), and Wiktionary's own entry on lleísmo. If you think some other term should be used, please give citations indicating that this is an accepted usage. Also the tag for lleísmo doesn't just say "rural Northern Spain" but also "Andes Mountains". I will change the latter text to read "Andes Mountains, Philippines" as I assume the Philippines is a seseo area. Benwing2 (talk) 00:22, 18 December 2022 (UTC)

Words ending in consonant+s

@Benwing2: For words ending in consonant+s (e. g. complots) or x (e. g. relax), the template assigns paroxytone stress, whereas they are oxytone. The problem lies on the line with the main accentuation rule:

if #syllables > 1 and rfind(word, "[^aeiouns#]#") or #syllables == 1 and rfind(word, "[aeiou]") then

But the rule for words ending in "n" or "s" is: not all such unaccented words are paroxytones, but only words with "n" or "s" not preceded by another consonant, said otherwise: preceded by a vowel. So how should the line be rewritten? Is the following variant correct?

if #syllables > 1 and (rfind(word, "[^aeiou#]#") or rfind(word, "[^aeiou][ns#]#") or #syllables == 1 and rfind(word, "[aeiou]") then

Burzuchius (talk) 10:39, 30 March 2022 (UTC)

• Fixed. Burzuchius (talk) 20:52, 12 April 2022 (UTC)

Is "dirección" not supposed to be /diɾekˈθjon/ instead of /diɾeɡˈθjon/?

In words like dirección, diccionario, and lección, shouldn't it be an unvoiced /k/ instead of a voiced /g/? The Spanish Wiktionary uses a /k/. Hvergi (talk) 20:42, 19 August 2022 (UTC)

See https://en.wikipedia.org/wiki/Spanish_phonology#Obstruents . Burzuchius (talk) 17:46, 20 August 2022 (UTC)

des-

I do not think that deshelar is pronounced differently from as if it were spelled *deselar, but it seems that at least some people pronounce deshue- differently from desue- (and perhaps deshie- from desie-). From Wikipedia: "In a number of varieties, including some American ones, a process parallel to the one distinguishing non-syllabic /i/ from consonantal /ʝ/ occurs for non-syllabic /u/ and a rare consonantal /w̝/. Near-minimal pairs include deshuesar [dezw̝eˈsaɾ] ('to debone') vs. desuello [deˈsweʎo] ('skinning')" (https://en.wikipedia.org/wiki/Spanish_orthography). Burzuchius (talk) 18:11, 18 December 2022 (UTC)

@Burzuchius Interesting, let me see if I can fix this. Benwing2 (talk) 01:05, 19 December 2022 (UTC)

Why does this show voiceless [s] before /l/?

Although the realization of /s/ as voiced [z] is far from categorical in any position, shouldn't this module display [z] rather than [s] before /l/ since it seems to consistently show voicing before other voiced consonants? Appendix B of the following article seems to show pretty high rates of voicing before a lateral: Sedó, Beatriz, Schmidt, Lauren B. and Willis, Erik W.. "Rethinking the phonological process of /s/ voicing assimilation in Spanish: An acoustic comparison of three regional varieties " Studies in Hispanic and Lusophone Linguistics, vol. 13, no. 1, 2020, pp. 167-217. https://doi.org/10.1515/shll-2020-2027 Urszag (talk) 14:39, 19 December 2022 (UTC)

Also, given the variability, why not transcribe the potentially voiced allophone of /s/ as [s̬] instead of [z], like how the voiced allophone of /θ/ is here transcribed as [θ̬] (e.g. in ˈaθ̬.me)?--Urszag (talk) 14:42, 19 December 2022 (UTC)

Hyphenation

@Benwing2: According to the RAE rules, hiatuses are not divided at the end of a line (except in prefixed and compound words). So, although the word "paella" has three syllables (pa-e-lla), according to the RAE rules there is only one place where it may be hyphenated: pae.lla. So, when showing hyphenation, sequences of vowels (no matter diphthongs or hiatuses, or even with an intervening "h") should not be separated: pa.ra.noia, ma.rihua.na.Burzuchius (talk) 16:08, 23 December 2022 (UTC)

I have changed the caption from "Hyphenation" to "Syllabification", since it is actually what the module shows. Burzuchius (talk) 18:35, 6 February 2023 (UTC)

/w̝/

We sometimes have [w̝] instead of [w]. I suppose when not postconsonantal. That's fine. My point is that this should be restricted to the phonetic form. The phonemic form should be always /w/. Same as /n/ corresponding to [n] or [n̪]. 92.218.236.121 15:11, 10 February 2024 (UTC)