Help talk:IPA/English

For requests for transcription, see Wikipedia:Reference desk/Language.

Iwan Rheon

We need help with his first name. If you can help, please see Talk:Iwan Rheon#English pronunciation. Kbb2 (ex. Mr KEBAB) (talk) 12:32, 11 June 2019 (UTC)

Incorporating audio pronunciation files?

Has there been discussion about the possibility of incorporating audio pronunciation MP3s? Of course that's an entire cottage industry, but it seems like it'd fit perfectly at home on WP. I don't even know if this suggestion is reasonable here. If not, I give any editor permission to delete this section (if that's even allowed; after all these years, I'm still rather unaware of WP etiquette) or leave a suggestion and if I end up back here I'll delete it myself. Bob Enyart, Denver KGOV radio host (talk) 21:57, 28 June 2019 (UTC)

What do you mean by "incorporating"? For what it's worth, templates like {{IPAc-en}} and {{IPA-es}}, which are used to show pronunciation in IPA in articles, already support accompanying a recorded sample, as in /ˌæləˈbæmə/ . If you're talking about adding audio samples on Help:IPA/English, I wouldn't support it because our guide is meant to be dialect-neutral. Nardog (talk) 17:40, 29 June 2019 (UTC)

Separate RP English and GA English?

It is predicted that RP and GA could split into two different languages. Although that might seem a bit “out there,” there are significant differences between the two. E.g., the derhotization and lengthening of vowels in RP vs the rhotization and shortening in GA, and also /t/, /d/ and others morphing into /ɾ/ in GA. There are so many differences, and I think at least these two (which could also be responsible for other dialects which came from those two main ones) deserve their own pages.

What do you guys think?

-Logan Sherwin Logan Sherwin (talk) 03:10, 19 July 2019 (UTC)

They're not actually that different. We've discussed having separate transcriptions in the past and the best approach we've come up with is this diaphonemic system that allows for a single transcription that encodes for multiple dialects. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 03:20, 19 July 2019 (UTC)

General American pronunciations of loanwords

How reliable are pronunciations such as /ˈkɔːʃiːtseɪ/ for Košice? The majority minority of GA speakers who have the cot-caught merger (which generally means that /ˈkɔːʃiːtseɪ/ would be rendered as [ˈkɑʃitseɪ], which I find unlikely), and our article General American tells us that foreign rounded mid back vowels are approximated with /oʊ/ in North America, unlike RP where both /əʊ/ and /ɒ/ are possible options.

Pinging IvanScrooge98 (talk · contribs). Do you usually check whether Americans really pronounce words as you suggest, or do you just go along with the dictionaries? Don't take it as an attack please. Kbb2 (ex. Mr KEBAB) (talk) 05:57, 17 August 2019 (UTC)

@Kbb2: actually I just check the dictionaries and keep the pronunciations that are possible according to English phonology, because sometimes it looks like they try to just approximate the original even when it wouldn’t sound natural in English (e.g. relying on AHD we would have something like /ˈkoʊʃɪtsɛ/, with /oʊ/ as you said but with a final /ɛ/ that couldn’t technically be there). Maybe that is the problem: sometimes you are not sure if it is the actual English pronunciation or just a way to come “closer” to the native one. イヴァンスクルージ九十八_（会話） 07:01, 17 August 2019 (UTC)

@IvanScrooge98: Cot-caught-merged speakers strongly prefer /oʊ/ over /ɑː/ for stressed foreign ⟨o⟩, probably because they consider the back [ɑː] and the central [äː] to be the same sound. The latter sounds absolutely nothing like ⟨o⟩ in e.g. Spanish. In fact, /ɑː/ is regularly used in the United States to approximate foreign ⟨a⟩, which is often central.

IMO all (or at least most) pronunciations of loanwords with stressed /ɔː/ (excluding the /ɔːr/ sequence) should be removed. See [1] and [2]. The pronunciation of foreign ⟨o⟩ as /ɔː/ is not only counter-intuitive in English (completely so in RP, not quite completely counter-intuitive in GA which has the lot-cloth split) but simply impossible for cot-caught-merged speakers who wouldn't use /ɑː/ but /oʊ/. Kbb2 (ex. Mr KEBAB) (talk) 07:24, 17 August 2019 (UTC)

@Kbb2: I understand. Or maybe we could add a note on this help for the speakers with the merger, unless /ɔː/ is also very uncommon for those with the split. イヴァンスクルージ九十八_（会話） 07:48, 17 August 2019 (UTC)

@IvanScrooge98: (The lot-cloth split is always present in those varieties of GA that don't merge cot with caught). I doubt that the way this works is that the speakers without the merger use /ɔː/ and those who have the merger use /oʊ/. Rather, I think that a source that claims that an American pronunciation of any given loanword contains a stressed /ɔː/ for orthographic ⟨o⟩ when not before /r/ and /ŋ, ɡ, θ, f, s/ is simply unreliable and should be dismissed (before /ŋ, ɡ, θ, f, s/ it should be checked in Longman Pronunciation Dictionary, which probably is a reliable source for American pronunciations). It doesn't list an actual American pronunciation, but rather tells Americans how to pronounce that word as closely to the original pronunciation as possible. That's different. Kbb2 (ex. Mr KEBAB) (talk) 10:14, 17 August 2019 (UTC)

@Kbb2: what I meant is whether /ɔː/ may be used by speakers with the split alongside /oʊ/ or is actually uncommon. But yes, we probably should check more reliable sources on this point. Thank you. イヴァンスクルージ九十八_（会話） 11:07, 17 August 2019 (UTC)

@IvanScrooge98: There's no GA without the lot-cloth split. You're mistaking the lot-cloth split with the cot-caught merger, which neutralizes the distinction between /ɑː/ and /ɔː/ and thus reverses the lot-cloth split by merging all instances of /ɔː/ with /ɑː/ (cot-caught-merged GA is an accent without contrastive THOUGHT, rather than one that doesn't have the lot-cloth split).

I find it implausible that speakers without the cot-caught merger would use /ɔː/ where those who have the merger use /oʊ/. I've never heard of such a distinction in AmE. Kbb2 (ex. Mr KEBAB) (talk) 12:30, 17 August 2019 (UTC)

When dealing with English pronunciations of places like this I prefer to see each different pronunciation given in each dictionary as a "data point" cos there sure as hell isn't a standard anglo approximation of "Košice" which everyone would agree on. I would consider such pronunciations reliable, but not necessarily exhaustive. I more often hear English speakers pronounce it with /i/ or /eɪ/ at the end than /ə/ but I suppose all three are possible, same for /ɔː/ x /oʊ/ x /ɒ/ in the first syllable. – filelakeshoe (t / c) 🐱 12:02, 17 August 2019 (UTC)

@Filelakeshoe: But there is a standard way of approximating foreign ⟨o⟩ in American English, and it's /oʊ/. GA doesn't have a contrastive /ɒ/ vowel, which is the same as /ɑː/ (which is the most common approximation of foreign ⟨a⟩). In BrE there's usually a choice between /ɒ/ and /oʊ/ - see [3] and [4]. Kbb2 (ex. Mr KEBAB) (talk)

BTW, just an idea here and feel free to dismiss it as my OR, but are pronunciations of European placenames in Webster necessarily contemporary approximations of local forms? There are Slovak immigrants in America since the 19th century, perhaps the /ˈkɔːʃiːtseɪ/ pronunciation has been in use since before the cot-caught merger was even a thing. For example I notice our article on Lodz gives the pronunciation /luːdʒ/ from Merriam-Webster which might reflect an archaic Polish pronunciation from before the /l/ → [w] shift completed, from the time when the Poles arrived to the US, perhaps... – filelakeshoe (t / c) 🐱 12:51, 17 August 2019 (UTC)

Interesting. My copy of Wells's Longman Pronunciation Dictionary gives /wʊdʒ; wuːtʃ/ for RP, /loʊdz; lɑːdz; wuːdʒ/ for GenAm, and [wutɕ] for Polish. Love —LiliCharlie (talk) 13:24, 17 August 2019 (UTC)

P.S. Maybe Merriam-Webster's /l/ doesn't "reflect an archaic Polish pronunciation from before the /l/ → [w] shift completed, from the time when the Poles arrived to the US." Article Łódź says: "In 1839, over 78% of the population was German," so I wonder if the /l/ of German was the decisive factor. Love —LiliCharlie (talk) 13:45, 17 August 2019 (UTC)

@LiliCharlie: That could've been the case. To this day I hear many Germans pronounce Polish ⟨ł⟩ as a plain /l/, which is a much better choice than /v/ (Standard German, as we know, doesn't have /w/). Kbb2 (ex. Mr KEBAB) (talk) 13:48, 17 August 2019 (UTC)

@Kbb2 and IvanScrooge98:: The typical pronunciation of foreign and especially first-time-heard foreign ⟨o⟩ in GenAm is indeed /oʊ/, regardless of the cot-caught merger. Foreign-name examples of this include Barbados, Bogotá, Dodoma, Mogadishu, Skopje, Xhosa, etc. (Notice that the current pronunciation on the page Mogadishu doesn't account for the GenAm pronunciation by listening to how Americans actually pronounce it here.) And by the way, I think most sources on the cot-caught merger agree not that The majority of GA speakers have it but rather that nearly a majority of GA speakers have it: so, in fact, a majority don't have it. Wolfdog (talk) 12:42, 17 August 2019 (UTC)

@Wolfdog: True, my bad. Kbb2 (ex. Mr KEBAB) (talk) 12:54, 17 August 2019 (UTC)

So from now on we should avoid listing these pronunciations, right? (With the exceptions above, possibly.) イヴァンスクルージ九十八_（会話） 13:44, 17 August 2019 (UTC)

Which pronunciations do you mean by "these pronunciations"? Wolfdog (talk) 15:26, 17 August 2019 (UTC)

@Wolfdog: the ones with /ɔː/ for ⟨o⟩. イヴァンスクルージ九十八_（会話） 15:45, 17 August 2019 (UTC)

What about /ɔːr/ and /ɔːl/ for ⟨or⟩ and ⟨ol⟩, wouldn't it be okay in those cases? Like in Olomouc and Córdoba. – filelakeshoe (t / c) 🐱 16:02, 17 August 2019 (UTC)

I don't think there's any need to avoid potentially helpful pronunciations (since there are plenty of exceptions to assumed rules of foreign pronunciations). We can certainly differentiate using the templates {{IPAc-en|US|...}} and {{IPAc-en|UK|...}}. As to Filelakeshoe's point, I'd say yes to the /ɔːr/ idea but I'd be cautious with /ɔːl/, since Americans tend to pronounce stressed ⟨ol⟩ as /oʊl/ (like in Angola, Kolkata, Mongolia, Poland, etc. off the top of my head). Wolfdog (talk) 16:18, 17 August 2019 (UTC)

That’s it. We should ascertain in which positions it is natural to have this sound and in which not, e.g. ⟨r⟩ normally affects the preceding consonant, so it would be perfectly fine to have /ɔːr/ in such transcriptions. イヴァンスクルージ九十八_（会話） 16:22, 17 August 2019 (UTC)

Does that mean that Amercan speakers who don't merge horse and hoarse have /ɔːr/ in those loanwords? Love —LiliCharlie (talk) 16:47, 17 August 2019 (UTC)

From the words listed at horse–hoarse merger, loanwords seem to go either way. "California" has the "horse" vowel and "Borneo" has the "hoarse" one. I suppose analogy to similar sounding words comes into play. – filelakeshoe (t / c) 🐱 17:47, 17 August 2019 (UTC)

@IvanScrooge98, Filelakeshoe, Wolfdog, and LiliCharlie: What about polysyllabic pronunciations in which most or all vowels are strong? For instance, /ˌɑːmɑːˈroʊneɪ/ instead of /ˌɑːməˈroʊneɪ/ for Amarone and /kɑːmˈpɑːnjɑː/ instead of /kɑːmˈpɑːniə/ for Campania strike me as non-English. It seems to me that many of the pronuciations of loanwords in the American Heritage Dictionary and, perhaps to a somewhat lesser extent, in MW are made up prescriptive nonsense that doesn't account for vowel reduction nor for the way Americans anglicize loanwords. Kbb2 (ex. Mr KEBAB) (talk) 10:48, 19 August 2019 (UTC)

@Kbb2: yes, in those cases we probably should avoid to transcribe them. Lately I have been trying to avoid them too. イヴァンスクルージ九十八_（会話） 10:54, 19 August 2019 (UTC)

I tend to agree with your prescriptive nonsense theory... or at least that there are still holdovers from the days when dictionaries subscribed to prescriptive nonsense. (One example that always springs to mind is how American dictionaries give the pronunciation of plantain as /ˈplæntɪn, -ən/ when I've never heard any Americans say this and the majority variant is certainly /plænˈteɪn/: a form that some, like Dictionary.com, don't even present as a possibility!) As for Amarone, the GenAm assumption probably would include the schwa unless the speaker was being very deliberate, while the Campania pronunciation would also include a schwa or two: /kɑːmˈpɑːnjə/ or /kəmˈpɑːnjə/ strike me as natural, though now listening to the 7 speakers of Youglish I hear an even larger diversity of (frankly, somewhat bizarre) pronunciations, which in addition to the variants we've given also includes variants like the stressed syllable being pronounced /peɪn/ or /pæn/ [pɛən]! So maybe there's nothing we can do for Campania. Wolfdog (talk) 11:20, 19 August 2019 (UTC)

@Wolfdog: I've fixed Campania using the RDPCE. It's almost funny how authors of transcriptions such as /kɑːmˈpɑːnjɑː/ don't realize that this pronunciation is almost as far away from the original as /kɑːmˈpɑːniə/. What's even more strange is that I've seen transcriptions of Dutch names in the American Heritage Dictionary that have /ɑː/ for both types of ⟨a⟩, even though length is essential (you'll be understood when you use [ɑː] for the front /aː/ and [a] or even [æ] for the back /ɑ/ but not when you get the length wrong). This is just weird. Kbb2 (ex. Mr KEBAB) (talk) 11:42, 19 August 2019 (UTC)

Yes, the same goes for German words like Bach, German [bäx], which is [bɑːk] for most English speakers, though short [-æ-] would be easier for German native speakers to understand. — I think that attempts at foreign pronunciations are rarely made in order to be more readily understood by foreigners, but to gain prestige among the people that surround you. Love —LiliCharlie (talk) 12:21, 19 August 2019 (UTC)

Yeah, the /ˈoʊloʊˌmoʊts/ someone just added to Olomouc is a bit strange. While it is true that Americans love butchering Czech by throwing unnecessary diphthongs everywhere (and it is butchering as it leads to the phonemes /o/ and /ou/ merging and no one understands them), surely an "English" pronunciation of a city name would be expected to adhere to English phonotactics which means the weak syllable in the middle here is definitely reduced. I have only recently been won over by the argument that having "English" pronunciations alongside the local ones is necessary and I might be returning to the other side of the fence, for cases where there is no established English usage. – filelakeshoe (t / c) 🐱 12:43, 19 August 2019 (UTC)

@Filelakeshoe: You can't expect Czech contrasts to be maintained in English. Americans approximate foreign ⟨o⟩ with /oʊ/ and it just so happens that Czech contrasts /o/ with /ou/, which is a non-starter for Americans (/ˈɑːləmoʊts/ is probably just as unacceptable to to cot-caught-merging Americans who don't have a phonemic /ɔː/ as it is to native speakers of Czech - another reason to be careful with /ɔː/ in loanwords). Kbb2 (ex. Mr KEBAB) (talk) 12:49, 19 August 2019 (UTC)

I am not expecting Czech contrasts to be maintained in English, I was referring to Americans not maintaining Czech contrasts while speaking Czech. I am still unconvinced that every "foreign ⟨o⟩" (in all 6000 odd dialects of "foreign") needs to be approximated as /oʊ/ even in weak syllables, less so that we should be telling people to say /ˈoʊloʊˌmoʊts/ but not /ˈɔːləmoʊts/ just to cater to the proclivities of people with a certain vowel merger. Please consider me opposed to the inclusion of these "English" pronunciations. Where there is no established usage, put the local pronunciation and let people decide how to approximate that in their dialect. This is why we have IPA help pages, with the "Nearest English equivalent" column. – filelakeshoe (t / c) 🐱 13:17, 19 August 2019 (UTC)

@Filelakeshoe: That's something else though. We're talking about English.

/ˈɔːləmoʊts/ is counter-intuitive to all Americans, not just those with the cot-caught merger. The default vowel for foreign ⟨o⟩ is /oʊ/, especially in syllables with primary stress. In less stressed syllables it can be /ə/ or perhaps /ɑː/. As Wolfdog has already confirmed, there's no such thing as speakers without the cot-caught merger using /ɔː/ where those with the merger use /oʊ/. There's no such dichotomy in American English.

If "Olomouc" were an English word stressed on the first syllable, the only possible pronunciations in General American would begin with /oʊ/ and /ɑː/. The lot-cloth split didn't affect /ɒ/ before /l/ - see English orthography#Combinations of vowel letters. This is why Mogadishu as /ˌmɔːɡəˈdiːʃuː, -ˈdɪʃuː/ isn't counterintuitive for Americans, but /ˈɔːləmoʊts/ is (I'd expect /ˈɔːləmoʊts/ to be spelled A(u)lomouts or something like that, compare Salzburg and Auschwitz).

LPD gives /ˈɒləmaʊts, ˈɒlɒ-/ as the RP variants and /ˈoʊloʊmaʊts/ as the GA variant. I imagine that the second /oʊ/ is weak, prone to be reduced to a schwa (so [ˈoʊɫɵmaʊts] ~ [ˈoʊɫəmaʊts]). Kbb2 (ex. Mr KEBAB) (talk) 08:18, 20 August 2019 (UTC)

Interesting, so LPD holds out against the Czech /au/ → /ou/ vowel shift, which completed in the 19th century. I think this entire thread proves my point that there is no established English pronunciation of Olomouc. Do Americans really pronounce "Auschwitz" with /ɔː/ at the beginning? Surely those Americans just don't know how the word is pronounced, and are taking wild guesses from reading the letters. – filelakeshoe (t / c) 🐱 08:39, 20 August 2019 (UTC)

John C. Wells, the author of the LPD, is aware that the current Czech pronunciation of ⟨ou⟩ is /ou/, and explicitly mentions that Olomouc is pronounced [ˈo lo mouts] (=[ˈo.lo.mouts]) in Czech — for the benefit of RP, GA, and other English speakers who wish to sound more Czech, but less English. (Note that according to LPD, English and Czech have different syllabifications: for RP and GA, the /l/ is the coda of the first syllable, for Czech, it is the onset of the second syllable. I don't think we should indicate syllable breaks, or only do so in exceptional cases, as this is a matter of considerable disagreement among authors of pronunciation dictionaries.) Love —LiliCharlie (talk) 10:29, 20 August 2019 (UTC)

Pronouncing the last syllable as /oʊ/ or /ou/ rather than /aʊ/ would not make one sound "more Czech and less English", it would make one sound more modern and less like they've quantum-leaped here from the 18th century. – filelakeshoe (t / c) 🐱 13:09, 20 August 2019 (UTC)

@Filelakeshoe: Remember that our /oʊ/ is just a diaphonemic notation. In England alone it can signify [əʊ], [əʉ], [əɨ], [ɐʊ], [ɐʉ], [ɐɨ], [oʊ], [oː] or other options. Many of these are significantly different from Czech /ou/ which shows no such variation (is [ˈɒləməɨts] or [ˈɒləmɐʊts] close to the original Czech pronunciation? The /t/ can also be pre-glottalized or glottalized, which isn't a thing in Czech) Because of that, I see no problem with transcribing the American pronunciation of "Olomouc" as /ˈoʊloʊmoʊts/ or /ˈoʊloʊmaʊts/, if the latter is an established pronunciation. Kbb2 (ex. Mr KEBAB) (talk) 13:25, 20 August 2019 (UTC)

@IvanScrooge98: /ˈrɑːdɔːm/ for Radom looks dubious to me. The sequence /ɔːm/ is very rare in General American and not an intuitive pronunciation of the orthographic ⟨om⟩. It seems to me that /ɔːm/ is just a Middle English variant of contemporary /oʊm/, as in "home" (ME /ˈhɔːm/), so not a sequence that would be used in rhotic English (except for a handful of words).

On the other hand, I'd expect e.g. the surname of Saša Ognenovski to be pronounced /ˌɔːɡnəˈnɔːfski/ or at least /ˌɑːɡnəˈnɔːfski/ in GA. /ˌoʊɡnəˈnoʊfski/ looks implausible to me, so /ɔː/, /ɑː/ and /oʊ/ seem to be in a kind of a complementary distribution in GA (at least as far as loanwords are concerned), with the first two being mostly confined to environments preceding (certain) consonant clusters (of course, /ɑː/ occurs in pretty much all positions when the spelling is ⟨a⟩, ⟨ah⟩, etc.). /ɔː/ can only occur before /ŋ, ɡ, θ, f, s/ and perhaps /n/, but not /m/.

I'd say let's just leave the UK pronunciation at Radom. It has an American-style /ɑː/ for the stressed ⟨a⟩, which is good enough. Kbb2 (ex. Mr KEBAB) (talk) 15:47, 25 August 2019 (UTC)

@Kbb2: it doesn’t to me, as it may in other cases, and it is reported by Oxford, which is definitely more reliable than other sources. However, considering what you say and since something with /oʊ/ is surely more likely, I wouldn’t really mind changing the pronunciation on the page, if you have references for a different one. イヴァンスクルージ九十八_（会話） 16:31, 25 August 2019 (UTC)

@IvanScrooge98: What's the reason you find it plausible, apart from the fact that it's listed in an Oxford dictionary? It's almost certainly a prescribed pronunciation, rather than a report of actual usage (as it kind of must be with names like this, Radom isn't Warsaw or Cracow). I still think it shouldn't be listed. Kbb2 (ex. Mr KEBAB) (talk) 16:56, 25 August 2019 (UTC)

@Kbb2: yeah, maybe by analogy with other nasals /n ŋ/ I felt it doesn’t sound implausible. Leave /ˈrɑːdɒm/. イヴァンスクルージ九十八_（会話） 18:20, 25 August 2019 (UTC)

@IvanScrooge98: Will do. And actually, /ɔːn/ is also a rare sequence and not necessarily an intuitive pronunciation of orthographic ⟨on⟩. /ɔːŋ/, on the other hand, seems to be natural for Americans.

IMO, the criteria for including a prescribed (rather than a confirmed, established) pronunciation of a loanword with /ɔː/ in a WP article should ultimately be the following: would a speaker of American English without the cot-caught merger pronounce /ɔː/ in that position if this were a native English word, or would the natural reflex be to use /oʊ/ or /ɑː/? If it's the latter, we shouldn't include it. Let's also not forget that /ɔː/ in New York and Philadelphia can be a centering diphthong [ɔə] or even [ʊə]. This /ɔː/ (as prescribed by dictionaries) obviously means General American [ɒː], rather than [ɔə ~ ʊə] in New York and Philadelphia or [ɑː] in some Inland Northern American English (where it contrasts with front unrounded [aː]). Not to mention that some North American English (notably Canadian) uses [ɒː] for the cot-caught merged vowel. So there's some overlap between back realizations of the (PALM-)LOT-THOUGHT vowel in cot-caught-merging dialects on one hand and the THOUGHT vowel in dialects without the cot-caught merger on the other. It's my suspicion that /ɑː/ and /ɔː/ are treated pretty much exactly as one vowel in loanwords, even in dialects that distinguish "cot" from "caught". They appear in complementary distribution and there's a preference for /ɑː/ wherever /oʊ/ (which is the primary choice) doesn't fit. It's obviously a different story when the spelling contains ⟨a⟩, as then there's a choice between /ɑː/ and /æ/ (or /ɑː/ and /ɛ/ before /r/), with /æ/ being rarer. Kbb2 (ex. Mr KEBAB) (talk) 08:11, 26 August 2019 (UTC)

What do you think, Nardog? Kbb2 (ex. Mr KEBAB) (talk) 08:47, 26 August 2019 (UTC)

@Kbb2: Banning specific derivations of vowels doesn't strike me as a good idea. Determining what counts as an incidence of such a derivation is not going to be easy; some borrowings may have assimilated to the English lexicon and genuinely be pronounced with /ɔː/. (It's also WP:CREEP-y.)

An alternative is to ban unnatural/artificial/prescriptive pronunciations more broadly, but this also doesn't seem practicable. The problem is that dictionaries rarely indicate which pronunciations are ones actually used by speakers and which are ones they recommend based on etymology or whatever. Perhaps the best we can do to prevent the latter from creeping in is to make it explicit at WP:PRON etc. that our documentation of pronunciations must be descriptive and that editors are encouraged to exercise editorial discretion when a sourced pronunciation is found to be leaning too much on the prescriptive side (in other words, just because an RS includes a transcription doesn't warrant its inclusion on WP). Nardog (talk) 03:45, 27 August 2019 (UTC)

This is a similar link that IvanScrooge98 & I discussed almost 4 months ago: https://en.wikipedia.org/wiki/Talk:Matterhorn#IPA_Pronunciation NKM1974 (talk) 18:34, 03 September 2019 (UTC)

If you need a native GA informant without cot-caught merger, feel free to ping me. For that matter, I wonder if it would make a useful subpage addition to WikiProject Language (or Linguistics, or English) to offer a page where users could voluntarily add themselves under headings listing various regional variants of pronunciation (or usage)? A corresponding Talk page there, could be used to poll users in a given group, when topics like this one arose. Mathglot (talk) 19:54, 6 September 2019 (UTC)

That's effectively a call for OR. Whenever one finds a pronunciation in a cited RS dubious, the only acceptable options are replacing it with another RS or removing it (whether BOLDly or based on a consensus). Nardog (talk) 20:12, 6 September 2019 (UTC)

Not when it's on a Talk page. It's the same idea as having pages, categories, and userboxes which list people who can translate between different languages. It's a volunteer resource to make Talk page discussion more effective. It doesn't change verifiability requirements in any way. Mathglot (talk) 23:48, 6 September 2019 (UTC)

Why is the letter "a" missing?

Why is the letter "a" missing from the chart? as in: "/ˈfaː.ðə/" from the line "(General Australian) enPR: fä'thə, IPA(key): /ˈfaː.ðə/". https://en.wiktionary.org/wiki/father#Pronunciation Misty MH (talk) 10:13, 22 August 2019 (UTC)

@Misty MH: We transcribe that vowel with ⟨ɑː⟩. Kbb2 (ex. Mr KEBAB) (talk) 10:14, 22 August 2019 (UTC)

Deprecate /ən, əm, əl/

I don't think we need /ən, əm, əl/ in this guide. Transcribing them with ⟨ə|n⟩, ⟨ə|m⟩ and ⟨ə|l⟩ is, IMO, good enough.

• These sequences are exactly the same phonemically. When /ən, əm, əl/ are analyzed as syllabic consonants /n̩, m̩, l̩/, they're often (though correct me if I'm wrong) analyzed as syllabic also in positions where an actual syllabic pronunciation isn't possible (IMO such an analysis is objectively less correct than analyzing them as /ən, əm, əl/, but that's just my opinion).
• We don't distinguish ⟨ən⟩, ⟨əm⟩ and ⟨əl⟩ from ⟨ə|n⟩, ⟨ə|m⟩ and ⟨ə|l⟩ in IPA (nor in the respelling system) but only in this guide and in the notes available to those readers that have access to the mouseover feature. This is utterly illogical, as we're effectively distinguishing and not distinguishing ⟨ən⟩, ⟨əm⟩ and ⟨əl⟩ from ⟨ə|n⟩, ⟨ə|m⟩ and ⟨ə|l⟩ at the same time. I could get behind writing "button" /ˈbʌtᵊn/, "rhytm" /ˈrɪðᵊm/ and "bottle" /ˈbɒtᵊl/ (though I wouldn't necessarily support /ˈbʌtn̩/, /ˈrɪðm̩/ and /ˈbɒtl̩/, /ˈbʌt(ə)n/, /ˈrɪð(ə)m/ and /ˈbɒt(ə)l/ or /ˈbʌtən/, /ˈrɪðəm/ and /ˈbɒtəl/ - the last option is truly awful, don't you think?).
• ...but there are two types of syllabic consonants: those commonly used by native speakers (which are recommended to non-native speakers), as in "button", "rhytm" and "bottle", and those that are only possibly syllabic (such as "children" and "barrel", but see below), with [ən, əm, əl] being the recommended pronunciation for non-native speakers (you can infer from that that such words are normally pronounced with [ən, əm, əl] in careful speech, which isn't necessarily the case with words like "button", "rhytm" and "bottle"). Dictionaries may or may not differentiate between those two groups, or mix them up while leaving out some words from the latter group. That is a problem in my view.
• The rules for writing ⟨ən⟩, ⟨əm⟩ and ⟨əl⟩ vs. ⟨ə|n⟩, ⟨ə|m⟩ and ⟨ə|l⟩ aren't fully straightforward and are different for each sequence. For instance, [m̩] can't appear after /t/ (unlike [n̩] and [l̩]) and only [l̩] can appear after /n/.
• Not only the rules themselves but also the exceptions to these rules are different; for instance, /ən/ is mandatorily full [ən] in "Southampton" (because /t/ is preceded by /p/) and /əl/ the orthographic ⟨l⟩ is varisyllabic after vowels - "mail" can have two syllables like "royal" or just one syllable (a pronunciation closer to spelling), which triggers different allophones of /eɪ/ in much of North America ([eɪ] before /ə/ and [eə] before /l/); also, the vocalized pronunciation of /l/ makes the disyllabic pronunciation mandatory - if I want to pronounce mail in the Estuary accent I have to say [ˈmeɪ̯ʊ], disyllabic like "royal" [ˈrɔɪ̯ʊ]. "Mail" is clearly reinterpreted as /ˈmeɪəl/ in accents (spoken by tens of millions of speakers) in which [ˈmɛːl] (= [ˈmɛːʊ]) is a possible pronunciation - see [5].
• /l/-vocalization makes words like "barrel" (which can be monosyllabic in traditional RP) mandatorily disyllabic.

/ᵊr/ (LPD transcription) in words like "Canberra" can be instead transcribed as ⟨ə|r⟩. I see no problem with that. Kbb2 (ex. Mr KEBAB) (talk) 15:56, 7 September 2019 (UTC)

Are you aware that ⟨ᵊ⟩ is the Handbook of the IPA's symbol #218 meaning "mid central vowel release"? Though ⟨ᵊ⟩ is frequently used in English phonology for an optional schwa it's probably better to stick to "official" IPA and not use ⟨ᵊ⟩ for a possible nasal or lateral release. Love —LiliCharlie (talk) 16:32, 7 September 2019 (UTC)

Of course. I'm not proposing that we use that symbol, though we could (per LPD and other dictionaries that use it). I'm proposing that we deprecate the distinction between ⟨ən⟩, ⟨əm⟩ and ⟨əl⟩ and ⟨ə|n⟩, ⟨ə|m⟩ and ⟨ə|l⟩ in favor of the latter. Kbb2 (ex. Mr KEBAB) (talk) 16:34, 7 September 2019 (UTC)

I don't think we're supposed to be distinguishing e.g. ⟨ən⟩ from ⟨ə|n⟩ any more than we're distinguishing ⟨ər⟩ from ⟨ə|r⟩. In both cases, the former is how we should be transcribing so that the mouseover features the correct example text. As far as I know, we don't ever ⟨ə|n⟩ except maybe in cases where there's a syllable boundary between them (like with fundamental, perhaps). The phonetic details for these sequences seems like a non-sequitor. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:23, 8 September 2019 (UTC)

I think you're taking this too seriously. The difference between əl etc. and ə|l etc. only exists in tooltips, which are only available to half the visitors, and most of the other half don't even realize they're there. So this strikes me as a problem that doesn't exist. I guess we could move the dipahoneme codes of the syllabic consonants to the bottom at Module:IPAc-en/phonemes, like we did /juː/ etc., but that's it. Perhaps /əl, ən, əm/ can be treated like /ɪl, ɪn, ɪm/ and /iə, uə/, which are essentially diaphonemes but are not distinct codes in IPAc-en.

I would strongly oppose using the superscript schwa. Like I said before, we'd have to agree on where exactly a syllabic realization can occur before doing that. Dictionaries aren't consistent: organ is considered to be able to have a syllabic /n/ by LPD but not CEPD, and only in RP by RDPCE. This is also one of the reasons I think taking the distinction between əl etc. and ə|l etc. seriously is doomed to failure. Nardog (talk) 15:37, 8 September 2019 (UTC)

Change /ɛər/ to /ɛːr/

Our article Help:IPA/Conventions for English lists six dictionaries that write the SQUARE vowel with ⟨ɛər⟩ (actually ⟨ɛə⟩ - from now on, I'll ignore the following approximant for the sake of simplicity). These are CEPD from 1930 (which now uses ⟨eə⟩, like most British dictionaries), the second edition of Oxford English Dictionary (more recent Oxford dictionaries use ⟨ɛː⟩ instead), Dictionary.com, A Pronouncing Dictionary of American English and the Macquarie Dictionary (but see Australian English phonology - a reformed phonemic orthography for AuE uses ⟨eː⟩). I think that none of our readers use the first one and not many use the second-to-last one, which leaves us with OED2, dictionary.com and the Macquarie Dictionary. That's three (or at best four) dictionaries that use ⟨ɛə⟩ for SQUARE. Most British dictionaries write it with ⟨eə⟩, but there's an increasing number of authors who opt to transcribe it with ⟨ɛː⟩, which is a more appropriate transcription (out of our ⟨ɛə⟩, British ⟨eə⟩ and the innovative ⟨ɛː⟩ the second symbol is the worst one - if you pronounce SQUARE like a literal [eə] in Britain it'll often be understood as NEAR. AFAIK, [eə] was never a part of RP, the first element of the diphthongal variety of SQUARE has always been open-mid or even lower - ⟨eə⟩ is just a simplified transcription, like ⟨r⟩).

Non-rhotic dialects that differentiate SQUARE from DRESS + /r/ often feature phonemic vowel length because of that. In England, Wales, Australia and South Africa the monophthongal pronunciation is already the norm. It's only in New Zealand (where the vowel tends to be the same as NEAR) and in some American and (mostly southern) English dialects in which the diphthong is still in use. By writing SQUARE with ⟨ɛːr⟩ we could help speakers of other dialects (as well as the non-natives) understand the nature of the phonemic contrast between SQUARE and DRESS + /r/. It's not diphthongization but length (also, in the environment where SQUARE contrasts with DRESS + /r/, the monophthongal realization is mandatory in all dialects - the diphthong is possible only before other consonants and in the word-final position, which can create pseudo-minimal pairs [I can't name any off the top of my head though, which is too bad]. Pseudo, because they clearly have different syllabification).

In rhotic dialects that contrast SQUARE with DRESS + /r/ (meaning: FACE + /r/ with DRESS + /r/) the realization of the former is predominantly monophthongal and close-mid [e], with any schwas being a mere phonetic detail - though correct me if I'm wrong here.

I guess that we use ⟨ɛə⟩ instead of ⟨eə⟩ because we write DRESS with ⟨ɛ⟩ rather than ⟨e⟩. All six dictionaries I've mentioned do not differentiate between the first element of SQUARE and DRESS, both are written with ⟨ɛ⟩. Help:IPA/Conventions for English currently lists 3 dictionaries (Shorter Oxford English Dictionary, OED3 and Lindsey's CUBE) that use ⟨ɛː⟩ for SQUARE, and all of them use ⟨ɛ⟩ for DRESS as well. So I guess that using either ⟨ɛə⟩ or ⟨ɛː⟩ for SQUARE when we write DRESS with ⟨ɛ⟩ is logical and consistent with the sources.

The 8th edition of Gimson's Pronunciation of English (2014) also uses ⟨ɛː⟩. EDIT: And so does the Routledge Dictionary of Pronunciation for Current English.

My point is that ⟨ɛː⟩ is already as recognizable as ⟨ɛə⟩. It's somewhat less similar to ⟨eə⟩ than ⟨ɛə⟩, but enough sources write SQUARE with ⟨ɛː⟩ to make this change unproblematic. It's an established transcription in a minority of sources, and the amount of sources using it will only increase, which likely won't be the case with ⟨ɛə⟩.

When we change ⟨ɛə⟩ to ⟨ɛː⟩ the speakers of GA-like accents can then simply ignore the length marks just as they ignore the distinction between ⟨iː⟩ and ⟨i⟩. In that regard, my proposal is consistent with the way our guide already works.

I wouldn't propose this if ⟨ɛə⟩ were an established symbol used by the majority of sources, but it's not. If we don't want to use ⟨eə⟩, then it's probably better to jump on the bandwagon and write ⟨ɛː⟩ instead. I don't think it will affect readability, not in 2019. If it was going to affect readability, I think that we'd already be receiving complaints about ⟨ɛə⟩, which is sort of a "mix" between ⟨eə⟩ and ⟨ɛː⟩.

I'm aware that we've already discussed this more than a year ago. Mach forgot to mention that Oxford dictionaries that are more recent than OED2 use ⟨ɛː⟩ and the CEPD has switched to ⟨eə⟩ decades ago. He did mention that, my bad. Also, Nardog, see [6] where JWL says that SQUARE "now seems to be increasingly regarded as more satisfactorily described as normally monophthongal [ɛː]. Accepting this view, I now normally write it as /ɛ/ with the economy of avoiding any length mark as unnecessary to distinguish it from /e/".

Again, the choice between ⟨ɛə⟩ and ⟨ɛː⟩ seems to be to a large extent arbitrary - so why not choose a symbol that's more correct as far as the actual pronunciation is concerned? Kbb2 (ex. Mr KEBAB) (talk) 18:23, 7 September 2019 (UTC)

I’d rather keep the diphthongal transcription. Another important aspect of phonetic symbols is their traditional use. We currently use Gimson’s system for transcribing English vowels (except for [ɛ]). It has a long tradition and a wide adoption, and it is still being used by the dedicated pronunciation dictionaries, which I would consider the best authority on such matters. When discussing Upton’s new system with signs like [ɛː], John Wells has said that “the supposed gains did not make up for the sacrifice of an agreed standard”, cf. IPA transcription systems for English.

Also, what about Upton’s other symbols? For sake of consistency, I think we should not introduce Upton’s [ɛː] without at least discussing [a], [əː] and [ʌɪ]. --mach 🙈🙉🙊 19:39, 7 September 2019 (UTC)

@J. 'mach' wust: The established (British) transcription of this vowel is ⟨eə⟩ (with ⟨e⟩ used for the starting point), rather than our ⟨ɛə⟩. Both denote centering diphthongs in IPA, but the latter seems to be almost as rare as ⟨ɛː⟩ in dictionaries/in the literature in general. That's the point: it's unlikely that the amount of sources using ⟨ɛə⟩ is going to increase. The amount of sources using ⟨ɛː⟩ is, on the other hand, almost definitely going to increase (Gimson's Pronunciation of English, one of the most recognizable works on English phonetics, has made that switch five years ago, also see English After RP and Vowel sounds of Received Pronunciation). Our readers that are familiar with the IPA are certainly familiar with all three transcriptions, and my hunch is that they already see ⟨ɛː⟩ more often than ⟨ɛə⟩ and both of them more rarely than ⟨eə⟩ (the latter part is self-evident, though).

We already deviate from Gimson's system in the way we transcribe SQUARE, GOAT (I'll make a separate thread about changing ⟨oʊ⟩ to Gimsonian ⟨əʊ⟩ when we reach a consensus here and in the discussion above, three open discussions is at least one too many) and with the way we treat the weak /u/, using it only in the prevocalic positions.

I'm not proposing that we implement Upton's system. I'm only proposing that we replace ⟨ɛə⟩ with ⟨ɛː⟩ (note that we already use Uptonian ⟨ɛ⟩). Only some sources that use ⟨ɛː⟩ also use ⟨a⟩, ⟨əː⟩ and ⟨ʌɪ⟩. I tend to disagree with using ⟨a⟩ for English TRAP as the vowel is fully front in most standard dialects, so it's better to use a special ⟨æ⟩ symbol for it (not to mention that raising it before nasals is very common in North America and Australia, in which case ⟨a⟩ is probably an objectively worse choice). In many languages other than English which have only one open vowel (most often transcribed with ⟨a⟩) its allophonic range is just too wide (some of these allophones would be interpreted as English /ʌ/ or, in extreme cases, /ə/). ⟨əː⟩ is logical but most sources don't use it and ⟨ʌɪ⟩ makes no sense from a phonetic perspective (and again - most sources don't use it). The difference between ⟨ɛː⟩ and ⟨a⟩, ⟨əː⟩ and ⟨ʌɪ⟩ is that while all of those aren't the established transcription used in the majority of sources, we write the last three sounds with ⟨æ⟩, ⟨ɜː⟩ and ⟨aɪ⟩ which are used by the majority of sources. ⟨ɛə⟩ isn't, and it's only somewhat more similar to ⟨eə⟩ than ⟨ɛː⟩.

I've just realized that the the Routledge Dictionary of Pronunciation for Current English (one of the three most recent pronunciation dictionaries for English) also uses ⟨ɛː⟩. I think that my theory about ⟨ɛː⟩ being already more recognizable than ⟨ɛə⟩ is true. Kbb2 (ex. Mr KEBAB) (talk) 03:05, 8 September 2019 (UTC)

@Kbb2: You are right that our using [ɛə] instead of Gimson’s [eə] is no automatic consequence from our using [ɛ] instead of Gimson’s [e]. If anything, I think we should use Gimson’s [eə].

Our system is already largely based on Gimson’s, and the small deviations only serve to make it more acceptable beyond England: We do not use simple [e] because of the American tradition of using that symbol for the FACE lexical set, and we retain [oʊ], thus preferring Jones’s older use over Gimson’s most idiosyncratic innovation (it is the only case where Gimson did not use a neighbour sign to Jones’s system). These two half-deviations (one sign in each pair still coincides with Gimson) do not change our deep roots in Gimsonian tradition.

Only some sources that use ⟨ɛː⟩ also use ⟨a⟩, ⟨əː⟩ and ⟨ʌɪ⟩. – What sources are these? In the sources gathered in Help:IPA/Conventions for English, there is only one that uses [ɛː] without fully adopting Upton’s system. It is the CUBE, which uses an idiosyncratic system devised by Geoff Lindsey. Unlike Gimson or Upton, Lindsey has never been adopted by any major dictionary. That is why I do not think that this major dictionary should use him as a reference. --mach 🙈🙉🙊 06:03, 8 September 2019 (UTC)

@J. 'mach' wust: My observation was that wherever SQUARE is written with ⟨ɛː⟩ or ⟨ɛə⟩, ⟨ɛ⟩ is used for DRESS. So there is a correlation. I'd argue that transcribing the former with ⟨eə⟩ and the latter with ⟨ɛ⟩, while not "wrong" per se and not unintelligible, will raise some eyebrows, though I don't think it'd take more than a couple of minutes (or a bit more than that) to get used to it. It's somewhat like transcribing the open vowels in German with ⟨ɑː⟩ and ⟨a⟩. It's not wrong, but nobody or hardly anybody does that (at least in modern literature). That's why I'm indifferent as far as the change from ⟨ɛə⟩ to ⟨eə⟩ is concerned.

I take your message as a no to ⟨ɛː⟩. Is that correct? Because we're not discussing changing ⟨ɛə⟩ to ⟨eə⟩, which is a closely related but still different topic. Changing ⟨ɛə⟩ to ⟨eə⟩ does make some sense, but to me it makes more sense to change ⟨ɛ⟩ and ⟨oʊ⟩ to ⟨e⟩ and ⟨əʊ⟩ along with it (as you can see I don't really buy that ⟨e⟩ could be a problematic notation) - but that, again, is a different topic (as is changing ⟨oʊ⟩ to ⟨əʊ⟩ regardless of the other two vowels).

The books I've mentioned. They also should be taken into consideration, IMO. Kbb2 (ex. Mr KEBAB) (talk) 15:49, 8 September 2019 (UTC)

Tl;dr but meh for ⟨ɛː⟩, strong oppose ⟨eə⟩. Introducing ⟨eə⟩ at this point disrupts the harmony where long vowels and centering diphthongs match the GA counterparts when ⟨ː⟩ or ⟨ə⟩ is removed. Replacing ⟨ɛ⟩ with ⟨e⟩ solves this, but that's obviously worse because ⟨e⟩ is a common symbol for FACE.

I don't buy the argument that if we were to introduce ⟨ɛː⟩ we should switch to Upton's scheme completely, because our system encompasses both RP and GA, and Cruttenden has only adopted ⟨a⟩ and ⟨ɛː⟩. Rather, it is when Wells, Roach, and Ashby (OALD)—or their successors—make a change that we'll need to seriously reconsider our set of symbols. But that doesn't at all seem imminent.

By the way, it is Windsor Lewis who devised ⟨eə⟩ for SQUARE, not Gimson. Gimson used ⟨ɛə⟩, probably following Jones, in his Introduction to the Pronunciation of English (1962), but used ⟨eə⟩ in EPD14 (1977) following Windsor Lewis's Concise Pronouncing Dictionary (1972). Windsor Lewis's recent use of ⟨ɛ⟩ for SQUARE doesn't prompt me to reconsider what I said last year, though, because that entails one uses ⟨e⟩ for DRESS. Nardog (talk) 16:37, 8 September 2019 (UTC)

You have convinced me that there is no need for changing our [ɛə] to [eə]. I still would not change it to [ɛː]. If I am not mistaken, then it is only Cruttenden who is using [ɛː] without fully adopting Upton’s system. However, Cruttenden’s solution is not used in any dictionary. I believe we should follow the dictionaries. --mach 🙈🙉🙊 22:00, 8 September 2019 (UTC)