Help talk:IPA/English/Archive 23

RfC: Should we acknowledge /ɜː/ as a marginal diaphoneme distinct from /ɜːr/?
There exist words such as föhn, Möbius, Peugeot, and pho, in which a vowel is pronounced as non-rhotic /ɜː/ in some accents, but not known to be pronounced as rhotic /ɜːr/ in rhotic accents. Currently, however, this key does not recognize /ɜː/ as a diaphoneme distinct from /ɜːr/ and IPAc-en automatically converts the input  to.

To accommodate this, should the key and template allow /ɜː/ to appear solely without /r/?

Articles affected by this proposal include, but are not limited to, American and British English pronunciation differences, Amuse-bouche, Beef bourguignon, Betelgeuse, Chartreuse (color), Foehn wind, Loess, Meunière sauce, August Ferdinand Möbius, Möbius strip, Peugeot, Pho, Richelieu, and Arnold Schoenberg.

See /Archive 21#Non-rhotic /ɜː/ for a preliminary discussion. Nardog (talk) 23:50, 5 June 2018 (UTC)

Pinging, who were involved in a discussion regarding this issue or in a previous RfC. – Nardog (talk) 23:50, 5 June 2018 (UTC)

Survey
Support. Hashed it out; the quibbles I might have had have been dispelled. — SMcCandlish ☏ ¢ 😼  01:12, 7 July 2018 (UTC)
 * Support adding /ɜː/. Currently we can only write e.g. ", ", which is confusing, and we should be able to write ", ". See /Archive 21#Non-rhotic /ɜː/ for my more detailed reasoning. Nardog (talk) 23:50, 5 June 2018 (UTC)
 * Abstain. Adding has a potential of confusing people just as much as transcribing pho as  does. I'd rather deal away with the whole diaphonemic system but that's obviously not the topic. Kbb2 (ex. Mr KEBAB) (talk) 00:09, 6 June 2018 (UTC)
 * Oppose for the same reasons as the previous discussion. Has anybody found any sources in this since then? — Æµ§œš¹  [lɛts b̥iː pʰəˈlaɪˀt] 02:18, 6 June 2018 (UTC)
 *  Potential support, iff we really need to . I have some level of skepticism about that, but if it's assuaged I would, of course, side with accuracy over convenience. I've detailed the concerns and quibbles in the "Discussion" section below.  — SMcCandlish ☏ ¢ 😼  05:02, 6 June 2018 (UTC)
 * Support Much more general, in my opinion, a valid IPA symbol should never be rejected or altered. &minus;Woodstone (talk) 05:25, 6 June 2018 (UTC)
 * Support flexibility. --mach &#x1f648;&#x1f649;&#x1f64a; 06:12, 6 June 2018 (UTC)
 * Support – not often required, but there are real examples as I've noted below. Peter coxhead (talk) 09:10, 6 June 2018 (UTC)
 * Support - I'm in favor of diaphoneme notation in general, and an R-less is no different.  (As it is, I still generally pronounce the weak vowel  [rounded, no off-glide] in minotaur, omit and polite differently from both  and  [rounded, with off-glide], and I'm still in my 30s.  But I digress.)  That said, I don't necessarily know how a lot words are normally pronounced outside my accent.  I can imagine  as a representative diaphoneme for a handful of words and utterances like duh, uh and um, but otherwise, I'm not clear which words would apply. - Gilgamesh (talk) 09:50, 6 June 2018 (UTC)
 * Support In words where even rhotic dialects don't pronounce the /r/, including it in the diaphonemic system is simply wrong. And it's certainly off-putting. --Macrakis (talk) 17:36, 6 June 2018 (UTC)
 * Support Some time ago I was puzzled at not being allowed to represent standard British pronunciations of the words listed above.  Wikipedia's insistence on adding an unsounded /r/ is very confusing, especially when the word is followed by a vowel where a linking r would be used for words that really do end with r.   Dbfirs  20:39, 6 June 2018 (UTC)
 * Weakly oppose Overall, this seems to me more conducive to controversy than clarity. I can already picture new or uninformed editors using this change as an excuse to argue that Birmingham is /ˈbɜːmɪŋəm/ rather than /ˈbɜːrmɪŋəm/, etc. Wolfdog (talk) 21:03, 6 June 2018 (UTC)
 * That's exactly my concern. Kbb2 (ex. Mr KEBAB) (talk) 21:06, 6 June 2018 (UTC)


 * Oppose - This is just the start of problems it will bring about. If you want to use, let's have separate RP and GA transcriptions. It's as simple as that. Kbb2 (ex. Mr KEBAB) (talk) 00:42, 7 June 2018 (UTC)
 * Ha! I can't believe my random hypothetical has been actually shown to have happened no less than a week ago! Great catch! Wolfdog (talk) 01:57, 7 June 2018 (UTC)
 * I'm sure Maczkopeti was simply correcting it in response to this RfC, not the other way around. Nardog (talk) 02:26, 7 June 2018 (UTC)
 * Makes sense. Didn't bother to notice the timestamps. Wolfdog (talk) 15:13, 9 June 2018 (UTC)


 * Oppose: This does not seem a proper diaphonemic symbol when American English is so inconsistent. It isn't clear what phoneme this symbol is supposed to translate to in GA. The symbol works best to describe non-rhotic dialects, but the removal of the r seems designed solely to signal that an American shouldn't pronounce an r in these words because there isn't one in the spelling. I sympathize with that sentiment, but often Goethe is pronounced as Gurta, Schoenberg as Shurnburg. I prefer Kbb2's suggestion of having separate RP and GA transcriptions. (I guess RP would have to stand in for Southern Hemisphere Englishes.) — Eru·tuon 03:29, 7 June 2018 (UTC)
 * Oppose: Although I did take precautionary measures to split from  where appropriate, I think the {{Search link|hastemplate:IPAc-en insource:/IPAc-en[^\}]*\| ?([\3ɜ]\:?|ɜː|\@\:)[\|\}]/|twentysomething}} cases where  is used to represent a non-rhotic foreign sound, it's UK-only, so it's going to be non-rhotic either way, and in cases like Goethe and Schoenberg, where it's also the US variant, it's rhotic, so  would be rather inappropriate. --maczkopeti (talk) 10:44, 7 June 2018 (UTC)


 * Support This is required for speakers whose accents have rhotic /ɜːr/ in most words but necessarily non-rhotic /ɜː/ in a few others. Indicating /ɜːr/ in all cases leads to mispronunciations. Love —LiliCharlie (talk) 00:33, 8 June 2018 (UTC)
 * Support: To put it simply, words like fauteuil (OED: ) and feuilleton cannot be accurately represented as  and  respectively, as even non-rhotic accents pronounce the /r/ when followed by a vowel. --maczkopeti (talk) 20:52, 9 June 2018 (UTC)

Discussion
Complex answer: I'm at least faintly skeptical that the average British person or any other English speaker actually uses /ɜ/ in any of the provided examples, rather than falling back to a more familiar /oʊ/ or something close to it, except when self-consciously over-enunciating for clarity (and sometimes perhaps for a socio-linguistic reason, often a temporary one predicated on who one is talking to/with). If I'm just provably wrong in this skepticism, and /ɜ/ really is a known norm in the UK [or a majority of it] for these words after all, then I agree, of course, with the proposed change – on the "convenience should not come at the cost of misinformation" principle. That rationale in detail: While we don't need use this to excess (per many previous discussions here and elsewhere in keeping IPA/English stuff as commonality-favoring as possible and not laden with "well, in part of the country ..." nit-picks), we should be able to do it accurately enough to avoid implying a /r/ sound that's not actually ever there. It'll just need an instruction to not use the character by itself to push a non-rhotic pronunciation of a word with an r in it which should thus have /ɜːr/ (because there are many British and related accents, some of which are rhotic and some not, with the later being innovations away from the former). We came to a previous conclusion that, for words like nurse, just rendering it /ɜːr/ is fine; someone natively rhotic is apt to not mentally hear that /r/ in the IPA anyway, for the same reason they don't in the plain-text string nurse ; it's fortuitous that IPA kept the same r caracter. In short: It would be better to a) avoid mistakenly implying /fɜːrn/ for föhn, at the cost of having to clean up stray instances of things like /nɜs/ as "British" [i.e., my local regional British] for nurse, than to b) automate in a way that makes /fɜn/ impossible output, just to make sure we always get /nɜrs/. This change will probably require more pronunciation gnoming, but that's the just price we pay for not auto-misrepresenting the British pronunciation of loans like föhn. , are we really sure about this pronunciation claim? In all such cases? In everyday, full-speed, casual speech? In a majority of British [or whatever range of] dialects? And how do we know on a particular-case basis? If you tell me Mötörhead is pronounced with this sound (twice!) I'm going to laugh, because I know for a fact that's not true. More seriously, I see real original research potential here, of people hyper-correcting towards what they think it should be rather than what it is. Do we have reliable sources, aside from dictionaries (which tend toward prescriptivism and over-simplification), for how most British people really say each of these things? I ask because I've seen this issue come up before many times. Two cases: So, let's be sure we need /ɜ/ before implementing it, since it does come at an intermittent "/nɜs/" cleanup cost. — SMcCandlish ☏ ¢ 😼  05:02, 6 June 2018 (UTC)
 * Example 1: People in New Mexico, especially Anglos and non-natively-Spanish-speaking Hispanics of central and northern New Mexican English, will often insist that chile (the regional variety of hot peppers, and by extension any non-local chili peppers) is pronounced toward the Spanish way, roughly /tʃileɪ/, to distinguish from Tex-Mex chili (a stew of red chile powder, beans, and ground beef), /tʃɪliː/. In actual practice, they only do this when trying to disambiguate, or when putting on airs as a "local" (a habit mostly found in urban and suburban not rural speakers). All the rest of the time, they pronounce both as /tʃɪliː/ – it's around 99% of the time, because the main usage phrases, "green chile", "red chile", and "chile con queso", have no ambiguity with chili-the-stew (technically, chili con carne). Yet they still have a strong subjective sense that /tʃileɪ/ is proper and is how they say it.
 * Example 2 is the well-known split in American English between /ɔːnt/ or /ɑːnt/ versus /ænt/, for aunt. It's a split which is confused and at times illusory. So many Americans are subjectively convinced that /ɔːnt/ or /ɑːnt/ is correct and that they use it, yet in recordings will actually strongly front the vowel without noticing, that some dialect surveys now ask separately about the word in isolation ("aunts and uncles", more likely to produce hyper-correction) and with a name ("ask Aunt Maggie", more likely to be uttered naturally without any socio-linguistic tweaking).


 * asks whether /ɜː/ is actually used in English. My experience, having lived in southern and midland Britain, is that there are words originally of foreign origin spelt without an "r" which I and others with a similar dialect pronounce using the same vowel as nurse, without any deliberate awareness that they are foreign words. A clear example is Peugeot as a kind of car, which I always hear pronounced as [pɜːʒoʊ] (or [pəːʒoʊ] as I would prefer to transcribe it following the OED now). Most of the other words in the list above are likely to be more familiar to people who know that they are foreign words or names, and so pronounce them at least sometimes with an awareness of this. I would (I think) always say [ɡɜːdəl] for Gödel, but would probably substitute other vowels in some of the other examples in informal speech. Loess and Schoenberg I often hear pronounced with disyllabic [oʊ ɛ], for example, although I would try to avoid doing this myself.
 * So do we need /ɜː/? Strictly, yes, since otherwise readers following our diaphonemic system who speak a rhotic dialect will pronounce our transcription /pɜːrʒoʊ/ in a way not intended. Peter coxhead (talk) 09:08, 6 June 2018 (UTC)
 * ... and I, from the north of England, agree with Peter on the use of /ɜː/, though I actually use /əː/ for nearly all of those words, and the OED uses /əː/ for quite a few of them.  Db</i><i style="color: #4fc;">f</i><i style="color: #6f6;">i</i><i style="color: #4e4;">r</i><i style="color: #4a4">s</i>  20:59, 6 June 2018 (UTC)
 * Let's not confuse phonemes with allophones. and  represent exactly the same thing. And actually, back in the day when Wells (or whoever that was) chose $\langle\rangle$ for this vowel it actually represented pretty much exactly the same sound as $\langle\rangle$. Kbb2 (ex. Mr KEBAB) (talk) 01:23, 7 June 2018 (UTC)
 * $\langleɜ\rangle$ was just an alternative symbol for $\langleə\rangle$ (or any central vowel higher than [ɐ]) until 1993 (see History of the IPA). The notation $\langleɜː\rangle$ for NURSE probably dates back to around the turn of the 20th century. Nardog (talk) 01:43, 7 June 2018 (UTC)
 * Yes, sorry to confuse the issue. I appreciate that Wikipedia uses  for the nurse vowel where I and the OED would use .  They don't sound quite the same in our IPA vowel chart with audio.   <i style="color: blue;">D</i><i style="color: #0cf;">b</i><i style="color: #4fc;">f</i><i style="color: #6f6;">i</i><i style="color: #4e4;">r</i><i style="color: #4a4">s</i>  06:35, 7 June 2018 (UTC)
 * You still seem to be, well, confusing phonemes and allophones. Like I said, both $\langleɜ\rangle$ and $\langleə\rangle$ referred to any sound between [ɨ] and [ɐ] until [ɘ] and the new value for [ɜ] were added in 1993, and /ɜː/ that we use is just a remainder from that period. We use it because it is the standard adopted by the authoritative English Pronouncing Dictionary and Longman Pronunciation Dictionary, as well as many advanced learner's dictionaries, and we do not claim its phonetic accuracy in any given accent (however it's pronounced, be it [əː], [ɚ], [ɘː], [ʌɾ], or [ɵɹ], our /ɜːr/ refers to the sound of NURSE—for [ ] vs. / /, see International Phonetic Alphabet, although I assume you're already familiar with it to some extent). (See for more.) Publications by Oxford University Press including OED have recently switched to /əː/, /ʌɪ/, etc., under the supervision of Clive Upton, but this scheme has met with not-so-positive response from fellow academics  and no other publisher has followed suit. Nardog (talk) 07:18, 7 June 2018 (UTC)
 * I can buy that (and also agree with you that our IPA symbol selection here isn't 100% ideal, but it seems "good enough" so I've never wanted to mess with it. >;-)  One of the reasons I asked about this outright is that, contra my first New Mexico example, there's another: people from the area tend to pronounce Spanish words (that have not been totally assimilated into English like "taco" and "macho") towards the Spanish vowel sounds.  E.g., I found it very jarring to hear Los Gatos, California, pronounced by people who live there as if it's Loss Gattoss.  So, it did occur that proximity to France and Germany might have an effect, but I was immediately skeptical the effect would persist island-wide, far away from the Channel.  I guess that question's still open.  I also asked because for the sound in question my own subjective experience is skewed. I never took a French class, but I did take a German one (and did not live near a French- or German-influenced area); consequently I lean very German-y on some of those words (even to /ø/) but in no way Frenchy on others, and this isn't a typical distribution.  So I thought, do we have any info, for the UK, one what really is a typical distribution?  — SMcCandlish ☏ ¢ 😼  10:34, 6 June 2018 (UTC)
 * I hope that you aren't suggesting that all people who pronounce aunt as /ɑ(ː)nt/ in North America are doing so artificially. At least in the area of New England that I am from, we pronounce aunt as /ɑ(ː)nt/ naturally. I knew of no other pronunciation until I first met (or first recalled meeting) some relatives from New York (back when I was a young child). Tharthan (talk) 12:16, 6 June 2018 (UTC)
 * Yes, /ɑ(ː)nt/ is consistently used by the people living around me in western New England, including even most of my young students. (I personally use it in a semi-consistent way that would make Tharthan frown, though I plead innocence, since I have many relatives from the NYC area who have influenced this pronunciation within my family circle). Wolfdog (talk) 21:36, 6 June 2018 (UTC)
 * Oh, of course; it's some variant of /ɑ(ː)/, /ɑː/, or /ɔ/ naturally and consistently in much of the US. The issue was that there's a subset of American English speakers who believe this is true of their variant but who nevertheless regularly use /æ/, especially when "aunt" is prefixed to a name. Dialect researchers have had to account for it – the perception of what is correct and even how they say it, among some speakers, didn't match the actual recorded data. Or in some cases, speakers consciously use two different pronunciations, in the same name-or-no-name pattern – I was in this latter group, and it's why I went looking into the matter way back when. It's parallel to New Mexicans' belief that they say /tʃileɪ/ being largely illusory or narrowly limited to a few circumstances. My concern was (maybe still is) that this effect might also apply to /ɜ/ (or /əː/ or /ɜː/, depending on sources) in the UK, in addition to the question of whether this Germanicization/Gallicisation effect is actually UK-wide, or just found in the south).  By now it's a moot point for this RfC, which is a snowball support,   [looks like I spoke too soon!]  and mainly a question for why to deploy /ɜ/ and on what evidence.  — SMcCandlish ☏ ¢ 😼  23:34, 6 June 2018 (UTC)
 * I'd like to give an American educator's perspective on the /ɜː/ issue without really knowing how it solves anything. Here are ways I've heard some of the "fuzzy" words said by Americans: Peugeot consistently with /uː/, Gödel and Goebbels usually with /ɜr/, Schoenberg with /ɜr/ or /oʊ/, Möbius consistently with /oʊ/ (I even sat through a whole play entitled this in NYC), pho with /oʊ/ (that makes me cringe) or /ʌ/, and milieu with /uː/ or /ʊ/. As you can see, American pronunciations are fairly all-over-the-place, while Brits can conceivably use /ɜː/ in place of every one of these vowels. Specifically in the German names Gödel and Goebbels (and perhaps others, like Schoenberg), most Americans I've heard do in fact use a rhotic sound! In the Vietnamese pho, on the other hand, Americans would find a rhotic sound completely bizarre.
 * I would like to say though that what might be most interesting is how such words are pronounced by people from the British Isles (and really specifically England) who still have rhotic accents. For example, does a West Country speaker say /ˈmɜrbiəs/, /ˈmoʊbiəs/, or truly this proposed new phoneme (which I've always assumed would be alien to the West Country phonological system) /ˈmɜːbiəs/? And how about Scots and Irish people: Do they side with the English, with the Americans, or with the beat of their own drummer? Wolfdog (talk) 21:33, 6 June 2018 (UTC)
 * Agree with all that, though the rhoticization is probably regional. I definitely hear it sometimes, especially from people from the eastern half of the country, and it sounds weird to me, a form of hyper-correction, like what English speakers have done to the word lingerie. It seems to happen more often the better-known the name/word is (e.g. more frequently to Goebbels than to Gödel). This may reflect some kind of shift in how people are approaching it (e.g. one's grandparents or great-grandparents from the WWII era may have had a different sense of how to handle ö than the average millenial does now).  I would guess that the broader cause of the US "do what you like" approach to (not just German) words and names that UK speakers might treat more consistently is that immigrants to the US have anglicized/Americanized their names in rather random or -random directions.  For a name like Groening, you have to ask the person how they say it (and Matt Groening, for example, insists on /eɪ/, which hasn't even been mentioned in this thread yet!).  On average, most Americans with Germanic oe names use /oʊ/. This "Ellis Island effect" is even more potent than one might think, because name spellings often shifted (Turberville, Paumgarden, Fingerhut from d'Urberville, Baumgarten, Vingerhoedt, etc.), with successive generations shifting pronunciations further to match the new orthographics.   regional dialect shifts have also had potent effects. It took me a long time to find some genealogical information on ancestors in Texas, because their Foster surname had been misheard by censustakers as Fauster or Fowster; it took Soundex searches to find them.  That brings me full circle to why I had doubts about the "it's /ɜ/ in the UK" assertion.  Because of its longer history, and despite its size, the UK has more dialectal variation than the US, and the variation is markedly deeper (plus also complicated by comparatively recent immigration waves from the Caribbean, South Asia, the Middle East, and Africa). If you include Ireland in the same dialect continuum, well, . >;-).  — SMcCandlish ☏ ¢ 😼  23:34, 6 June 2018 (UTC)
 * I also have had the same experience that, of the rhotacized form, I "definitely hear it sometimes, especially from people from the eastern half of the country, and it sounds weird to me". I should've made it clear that since these are all rare and learned words/names, I've only picked up their pronunciations later in life by imitating someone or (if lucky) a small group of speakers. In fact, I couldn't readily tell you my own pronunciations for them with perfect certainty, though this is my best guess: P[u]geot, G[œ]del and G[œ]bbels (maybe, in fact, [œ˞], though the strongly rhotacized form, [əɹ], I agree, sounds weird), Sch[oʊ>əɹ]nberg, M[oʊ]bius, ph[ʌ], and mil[ju>ʊ]. Incidentally,, what do you think would be your more Western U.S. pronunciations of these same words? (We can take this to my talk page if you feel we're going to off the main track here.) Wolfdog (talk) 02:26, 7 June 2018 (UTC)
 * Yeah, I'll user-space it, since I'm not a typical sample for an American (formative years in England, etc.)  — SMcCandlish ☏ ¢ 😼  12:02, 7 June 2018 (UTC)
 * Youglish is a pretty handy tool to gauge how people actually pronounce certain words. And dictionaries do indeed transcribe many words with /ɜː/, not only for RP but sometimes for General American too. To quote my original post:
 * In rhotic accents, such a vowel is pronounced as either:
 * another vowel such as /uː/ (Betelgeuse) or /ʌ/ (pho), sometimes even violating the phonotactics;
 * the marginal vowel /œ/ (found in American dictionaries, though I'm not sure how much this is meant to be descriptive of English speakers' actual production as opposed to the pronunciation in the original language);
 * a lengthened schwa-like vowel, just like the non-rhotic NURSE; or
 * the usual rhotic NURSE vowel, as in many speakers' pronunciation of Goethe and hors d'oeuvre.
 * Cambridge English Pronouncing Dictionary, 18th ed., includes at least 42 instances of case #3 (accoucheuse, adieu, Auteuil, Beaulieu (in France), boeuf bourguignon, Böhm, Böll, Châteauneuf-du-Pape, coiffeuse, Creuse, Deneuve, Depardieu, Des Voeux, Dieu et mon droit, Dönges, douloureux, émeute, feuilleton, Goebbels, Goethe, Göteborg, götterdämmerung, Greuze, jeu, jeunesse dorée, Köchel, masseuse, milieu, millefeuille, mitrailleuse, Mönchen-Gladbach, Montreux, Norrköping, oeuvre, Peugeot, pot-au-feu, prie-dieu, roman fleuve, Seurat, soixante-neuf, van Rompuy, Villeneuve) and 17 of case #1 (Bayeux, berceuse, Betelgeuse, chacun à son goût, chanteuse, chartreuse, cordon bleu, danseuse, fauteuil, föhn/foehn, Loeb, meunière, Meuse, Neuchâtel, Neufchâtel, Richelieu, Schrödinger), and Longman Pronunciation Dictionary, 3rd ed., includes at least 52 instances of #1 (Bayeux, berceuse, Betelgeuse, boehmite, boeuf, Böhm, chacun à son goût, chanteuse, chartreuse, Châteauneuf-du-Pape, cordon bleu, danseuse, Dieu et mon droit, Eupen, faites vos jeux, faute de mieux, föhn/foehn, Gödel, Goebbels, Goethe, Göttingen, Hebei, Henan, Hoechst, jeu(x) d'esprit, jeunesse dorée, Königsberg, masseuse, meunière, Meuse, milieu, Möbius, Montesquieu, Monteux, Montreux, Neuchâtel, Norrköping, oeil-de-boeuf, oeuvre, Peugeot, pot-au-feu, prie-dieu, Richelieu, roman fleuve, sauve qui peut, Schönberg/Schoenberg, Schrödinger, Seurat, soixante-neuf, Veuve, Villeneuve, Zhejiang) , counting only the first variants in each accent.
 * Listen, for example, to . There's no denying that non-rhotic speakers do use /ɜː/ in these environments. Nardog (talk) 01:39, 7 June 2018 (UTC)
 * I'm not sure what you mean by giving all these links and then saying that "There's no denying that non-rhotic speakers do use /ɜː/ in these environments". OK, but most of the speakers are rhotic and probably American. Listening to the first few American pronunciations of adieu, all the speakers say it homophonously with ado /əˈduː/. Where's the /ɜː/ in that? The same with Betelegeuse and some others. Here's what I hear from the first five "Schoenberg" audio clips: [əɹ(ə)], [oʊ], [oʊ], [əɹ], and [əɹ]. I'm not sure if I've missed your point or you did, ha. Wolfdog (talk) 02:14, 7 June 2018 (UTC)
 * I was just addressing SMcCandlish's original question, nothing more. The links refer to specific videos (also note many of what are transcribed in CC as adieu actually are [without further] ado). Nardog (talk) 06:52, 7 June 2018 (UTC)
 * Some more analysis of this sort could eventually address my rather diffuse concerns for the most part, perhaps even the one about whether we're taking southern English to be "British" and wrongly imposing its norms more broadly, but it would take a more in-depth review of data like this to be certain on that point. [Aside: This thread is an example, worth saving, of how WP:NOR doesn't apply to talk pages and the decisions that result from consensus on them. We frequently and permissibly do direct OR, including all four of the letters in WP:AEIS (analysis, evaluation, interpretation, and synthesis), of source material in the course of examining it and making internal governance decisions based on such material and on other, more WP-specific, considerations. It never ceases to amaze me how many editors, many of whom have been around long enough to know better, believe that NOR prohibits what we do routinely; it's related to the problem of people insisting we must cite sources in the MoS and other guidelines or in the documentation of templates and help pages as if they're articles.]  — SMcCandlish ☏ ¢ 😼  02:17, 7 June 2018 (UTC)
 * The pronunciation dictionaries I cited are the main, reliable sources for the proposal. The Routledge (formerly Oxford) Dictionary of Pronunciation for Current English, edited by the same people behind OED's pronunciations (Upton & Kretzschmar), also transcribe at least 43 words with /əː/ (= /ɜː/) for the British pronunciation, again counting only the first variants. The video links I provided are auxiliary, non-reliable sources for us to empirically confirm that speakers do indeed use /ɜː/ for those words. Nardog (talk) 11:30, 7 June 2018 (UTC)
 * I'm actually not a fan of using dictionaries for this sort of thing if more reliable linguistic sources are available (journal studies and the like, based on large corpora), because dictionaries are prescriptive and generally based on a dominant or "prestige" dialect, like Midwestern in the US and "received" pronunciation in the UK. They're a form of mild PoV pushing when it comes to pronunciation material, though basically out of being one-size-fits-all not due to having an agenda (American Heritage excepted – it has an agenda). For the US there's an enormous multi-volume atlas of this sort of linguistic data, but it costs about $1000 or so; something one would probably have to consult at a university library's reference room. Given that, I guess I'm okay with relying on dictionaries for the short term, but they're basically challengeable at any time by better sources.  — SMcCandlish ☏ ¢ 😼  11:46, 7 June 2018 (UTC)
 * I'm actually not a fan of using dictionaries for this sort of thing if more reliable linguistic sources are available (journal studies and the like, based on large corpora), because dictionaries are prescriptive and generally based on a dominant or "prestige" dialect, like Midwestern in the US and "received" pronunciation in the UK. They're a form of mild PoV pushing when it comes to pronunciation material, though basically out of being one-size-fits-all not due to having an agenda (American Heritage excepted – it has an agenda). For the US there's an enormous multi-volume atlas of this sort of linguistic data, but it costs about $1000 or so; something one would probably have to consult at a university library's reference room. Given that, I guess I'm okay with relying on dictionaries for the short term, but they're basically challengeable at any time by better sources.  — SMcCandlish ☏ ¢ 😼  11:46, 7 June 2018 (UTC)

Should we decide to introduce non-rhotic /ɜː/ I propose to input it as  in template. Would  be a suitable mouseover text? Love —LiliCharlie (talk) 01:34, 7 June 2018 (UTC)
 * Let's worry about that after the consensus is made to actually add it. You're counting your chickens before they're hatched. Nardog (talk) 01:46, 7 June 2018 (UTC)

Your argument sounds like a false dichotomy to me. People not getting WP:RHOTIC is nothing new; transcriptions violating our diaphonemic principle are already being instated left and right, and it's not like adding or not adding /ɜː/ can considerably spur or deter it. And last I checked (around the time of my original post), the number of the uses of,  , etc. where it should have been /ɜːr/ was only a few dozens (looks like Maczkopeti has fixed them so I can't get an accurate number). Nardog (talk) 01:57, 7 June 2018 (UTC)
 * Except that the reasonable concern expressed by 3 or more of us is that it predictably will actually spur this problem for a particular sort of case. I'm not sure that's in any way "fatal" to the proposal because the cleanup isn't dreadfully hard, but let's not try to sweep it under the rug. The entire nature of the proposal is that the character (which ever one we pick for the limited orthography of IPA/English) isn't usable by itself with the template yet, so how often it's abused now (e.g. by someone manually inserting it in untemplated attempts at IPA) is essentially irrelevant; it'll be orders of magnitude more easy to do so if the change is adopted. It seems like arguing that cyanide should be legal to sell at the corner store because it can't really be that dangerous judging from the number of annual cyanide poisonings right now. :-)  — SMcCandlish ☏ ¢ 😼  02:20, 7 June 2018 (UTC)
 * I don't know if this is throwing a wrench in the whole general bend of the discussion here (yay for mixed metaphors), but would just some other symbol be preferable? ...for example /œː/ as used in some of the dictionaries we've inspected? Wolfdog (talk) 02:33, 7 June 2018 (UTC)
 * Quite possibly. I think we've seen at least 3 symbols so far being used for the same phoneme in the same words, depending on source, and one of them was more similar to the one under discussion, so I'd go with that one.  — SMcCandlish ☏ ¢ 😼  02:38, 7 June 2018 (UTC)
 * AFAIK no dictionary transcribes any English word with $\langleœː\rangle$. I assume you meant $\langleœ\rangle$, which is seen in American dictionaries, but it's not nearly the same thing as the /ɜː/ we're talking about here (/ɜː/ encompasses beyond just front rounded vowels, as in pho ). I'd oppose adding it as it has even less evidence of its status as a distinct phoneme and seems to be more of a shorthand for "a front rounded vowel in the original language" (like I said in January). We've vetoed the proposal to add /ɑ̃ː, ɔ̃ː, ɜ̃ː/ for similar reasons.
 * SMcCandlish, by "3 symbols" are you talking about $\langleɜː\rangle$, $\langleəː\rangle$, and $\langleœ\rangle$? If so $\langleɜː\rangle$ and $\langleəː\rangle$ are literally the same thing, as I said above, while $\langleœ\rangle$ is a totally different thing. Nardog (talk) 03:06, 7 June 2018 (UTC)
 * Yes, I know. What I mean is that the glyphs are not the same, so the template code can distinguish them.  So, if we're using /ɜː/ in /ɜːr/, we could use /əː/ for stand-alone use, and this would thwart some attempts to convert /ɜːr/ incorrectly to /ɜː/, just to mimic local non-rhotic pronounciation, in names/words like Birmingham.  — SMcCandlish ☏ ¢ 😼  11:39, 7 June 2018 (UTC)

Can't you say the same thing for any marginal diaphoneme that we already have? For instance, /x/ is probably non-contrastive for many speakers, and CEPD and LPD both show bon vivant as /ˌbɑːn.viːˈvɑːnt/ as the first variant in GA. The intended use of /ɜː/, should we agree to allow it, is to write like ", ", or when the US/rhotic variant is unknown, just "", "", "", or so on. We should be using /ɜːr/ only when we know for a fact it's pronounced as [ɜ(ː)r] in rhotic accents, as in, like you say, Goethe and Schoenberg, or otherwise wouldn't it constitute more of OR if we were writing /ɜːr/ when we don't know if it's pronounced as [ɜ(ː)r] in a rhotic accent but we know for a fact it's pronounced as [ɜː] in a non-rhotic accent? Nardog (talk) 06:01, 7 June 2018 (UTC)

Perhaps I didn't do a good enough job explaining what the proposal was. The proposal isn't to signal that an American shouldn't pronounce an r in these words because there isn't one in the spelling, or to use /ɜː/ in cases like Goethe and Schoenberg, where it's also the US variant. It is a proposal to make it possible to use /ɜː/ for a vowel that is pronounced as [ɜː] in non-rhotic accents, but that is pronounced as some other vowel in rhotic accents, or that we can't say for sure is pronounced as [ɜ(ː)r] in rhotic accents. Goethe and Schoenberg must be transcribed with /ɜːr/ anyway because they are for sure pronounced as [ɜ(ː)r] in rhotic accents. Nardog (talk) 11:04, 7 June 2018 (UTC)
 * Note in the printed editions and other electronic editions of the MWCD it is actualy goetite \ˈgə(r)-ˌtīt\. So it seems they use "ə(r)" for any /œ~ø/ foreign sounds impliying that it can be pronounced both ways with and without "r", or, if using the IPA, /ɜːr/ and /ʌ/ respectively.--Lüboslóv Yęzýkin (talk) 19:32, 8 June 2018 (UTC)
 * That was my impression, that the proposed would be a non-rhotic-only diaphoneme, so it doesn't really change my opinion. The rest of the symbols are meant to cross the rhoticity divide, including the nasal vowels and, which you mentioned above. Adding a symbol unlike the others muddies up the system. However, I wouldn't like seeing the RP or non-rhotic pronunciation (I don't know how Australians or New Zealanders say these words) of foehn transcribed as  either, so I do understand the impulse. — Eru·tuon 04:05, 9 June 2018 (UTC)

I don't think many people would be amused by your assertion that it's UK-only, so it's going to be non-rhotic either way. If we equated the UK prefix with RP, that would be true, but currently our Help:IPA/English key doesn't present itself that way. Nardog (talk) 11:08, 7 June 2018 (UTC)

What do we think of the suggestion of simply avoiding the diaphonemic system in the relevant situations? Wolfdog (talk) 14:44, 7 June 2018 (UTC)
 * If I understand the proposal correctly, it's not creating a new diaphoneme that is pronounced one way by non-rhotic and another way by rhotic speakers so that we can transcribe these words with a single transcription. Rather, it's to provide an opportunity to give two transcriptions where one is the nurse vowel and the other is another vowel that isn't the nurse vowel and is spoken by rhotic speakers.
 * This is rather convoluted and I think belies a misunderstanding of what our system is presenting. We chose $⟨ɜr⟩$ to represent the vowel of nurse because of the correspondence between rhotic and non-rhotic accents. But that decision doesn't mean we are claiming with every transcription using $⟨ɜr⟩$ that there must be a dialectal concordance. It's simply how we represent the nurse vowel.
 * We've already done away with using an extra symbol to mark a difference of incidence in one transcription. I don't think it's a good idea to have two symbols that mean the exact same vowel diaphoneme just because one set of dialects favors that vowel in certain words. It would, in effect, be putting a non-diaphoneme in a diaphonemic system.
 * Moreover, it is by no means the only case of this sort of thing happening. By means of analogy, we could also have a separate symbol representing the vowel of ash that appears in loanwords from Romance languages, since British speakers tend to use it instead of the vowel of father? I don't think that would be a good idea either.— Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt] 21:55, 7 June 2018 (UTC)
 * All this is quite true from a North American point of view. However there seem to be accents in northern Britain, and maybe also elsewhere, that exhibit rhotic /ɜːr/ in most cases but also have a set of words in which /ɜː/ may have no r-colouring. Giving only /ɜːr/ is definitely inadequate for speakers of those accents as it leads to mispronunciations. (Note that the letter $\langler\rangle$ does not always identify rhotic /ɜːr/; colonel and kernel are both .) Love —LiliCharlie (talk) 00:21, 8 June 2018 (UTC)
 * Is this something that we can verify from dictionaries? — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt] 17:39, 8 June 2018 (UTC)
 * Sure. Collins indicates the difference, e.g. /bɜːʳn/ for burn but /fɜːn/ for foehn. Love —LiliCharlie (talk) 20:24, 8 June 2018 (UTC)
 * Sorry, but that's a failed verification. The burn example is not labeled for region, so it's presumably diaphonemic, while the foehn one is specifically labeled as British. As we can see from burglarize, Collins has non-diaphonemic transcriptions for specifically American and British pronunciations; they are using superscript r to indicate a US/UK (or rhotic/non-rhotic) variation with one diaphonemic transcription. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt] 21:05, 8 June 2018 (UTC)
 * I see. The Macmillan Dictionary of British English has entries such as /fɜː(r)/ fur vs. /ˈmiːljɜː/ milieu and /bɜː(r)n/ burn vs.... Well, unfortunately a search for foehn/föhn/fohn yields no result in their online version, so my example of pre-consonantal /ɜː/ is /ˈɜːvrə/ oeuvre. (Their American English dictionary gives the following pronunciations: /fɜr/, /milˈju/. /bɜrn/, and /ˈʊvrə/.) Love —LiliCharlie (talk) 01:39, 9 June 2018 (UTC)
 * Are we to understand Macmillan's $⟨ɜː(r)⟩$ as meaning for British non-rhotic accents and  for British rhotic accents?  I mean, as opposed to them putting into practice what is being proposed here. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt]  14:57, 9 June 2018 (UTC)
 * Yes we are. Macmillan's BE dictionary always shows where /r/ is pronounced in rhotic accents of the British Isles, e.g. /klɑː(r)k/ clerk (AE /klɜrk/). (Maybe this is no coincidence. After all, the founders of Macmillan Publishers were from rhotic Scotland.) Love —LiliCharlie (talk) 15:38, 9 June 2018 (UTC)
 * I'm not sure if I get how you know it's one and not the other. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt] 17:56, 9 June 2018 (UTC)


 * The British English edition of the Macmillan Dictionary has "entries which ... show British pronunciation" while the American English edition shows American pronunciations. Neither indicates pronunciations used on the other continent. Love —LiliCharlie (talk) 20:22, 9 June 2018 (UTC)
 * I'm still not clear on whether we have some confirmation that is used for rhotic (as well as the non-rhotic) British accents in a word like "oeuvre". And I mean in the Macmillan Dictionary or otherwise. Wolfdog (talk) 21:10, 9 June 2018 (UTC)
 * Yeah, your answer doesn't really clarify how we can be sure that the vowel in question is supposed to be a non-rhotic vowel pronounced by rhotic British speakers. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt] 23:03, 9 June 2018 (UTC)
 * What if the dictionary makers confirm by email that $⟨(r)⟩$ $⟨ɜː⟩$ (as opposed to $⟨ɜː(r)⟩$) in Macmillan's BE dictionary is "supposed to be a non-rhotic vowel pronounced by rhotic British speakers"? Love —LiliCharlie (talk) 08:48, 10 June 2018 (UTC)
 * It could be the case that Scottish people pronounce words such as milieu as, akin to for idea. But I don't have a source for that. All I know is that vowel length works differently in Scotland in comparison with other places and it may be a bit unrealistic to expect Scottish people to produce the same kind of  as Brits living down south. Again, this is all speculation. Kbb2 (ex. Mr KEBAB) (talk) 12:48, 10 June 2018 (UTC)
 * Just listening to some Brits saying Möbius here and some say /ɜː/ while others certainly say /oʊ/ [əʊ]. And, interestingly, a Scottish speaker in video 5 out of 8 clearly says /oʊ/. And, also for fun, I see that of the first 10 videos on the site for Americans pronouncing "Goethe" (actually 13, because three were clearly not native speakers of American English and one was blatantly British), six pronounce the name with a rhotic /r/ sound (one, even pronouncing it /ɛər/ rather than /ɜr/) and four don't. Again, it seems to me that the diaphonemic system can't fix this problem of variability, regardless of what new phonemes we "invent". Wolfdog (talk) 14:23, 10 June 2018 (UTC)
 * Huh, I didn't hear any cases of in those videos of Möbius, only . Which ones did you identify as ? Apparently Möbius strip is inaccurate when it indicates Merbius is the only pronunciation in the UK. — Eru·tuon 05:14, 11 June 2018 (UTC)
 * I agreed when I first heard them but worry that some of this may be our own bias towards American sounds. That said, you may be right after all, . At least to my American ears, the distinction between British /oʊ/ and British /ɜr/ is very very subtle in quickly-spoken words, where either can be heard as something like [ə(ː)]. Wolfdog (talk) 21:27, 11 June 2018 (UTC)
 * I agree. It's clear how similar they are in Geoff Lindsey's transcription system: . — Eru·tuon 22:23, 11 June 2018 (UTC)
 * A random example that comes to mind is [ˈfləɨtɪŋ] versus [ˈfləːtɪŋ]. One might be floating and the other flirting, but the two are damn close! Wolfdog (talk) 22:31, 11 June 2018 (UTC)

FYI: Move proposal of IPA chart for English dialects
A user has started a move proposal of International Phonetic Alphabet chart for English dialects to Help:IPA/English dialects. Nardog (talk) 00:33, 1 August 2018 (UTC)

Applying the diaphonemic principle to local pronunciations
These edits by Kbb2 got me thinking: Should we apply the diaphonemic principle to pronunciations labeled as "local" too?

If Bury is pronounced as [ˈbʊri] in Bury, it doesn't help much to say the alternative local pronunciation is when the fact that the dialect spoken there lacks the foot–strut split is not immediately available in the lead (which it shouldn't), does it? Writing " locally also " allows those who have the split to identify the pronunciation it refers to upon hearing it and to imitate it, which "" doesn't.

I know some people might be going to say IPAc-en shouldn't be used for local pronunciations in the first place, but I think in these cases it's better than any alternative MOS:PRON recommends. IPA-endia doesn't work because the chart it links to doesn't cover local dialects. IPA-all would falsely say that [r] is a trill. Linking the notation to Brummie dialect or Manchester dialect ad hoc wouldn't help either because those articles don't define the value of each phoneme. Whenever the purpose of a notation is to illustrate a difference that can be explained using our diaphonemes, both using IPAc-en and not applying the diaphonemic principle despite it seem reasonable to me. Nardog (talk) 08:07, 9 August 2018 (UTC)
 * I wouldn't use IPAc-en for local pronunciations like this one. It guarantees inconsistencies - think of non-rhotic dialects, dialects with a separate FORCE vowel, Canadian raising which may be analyzed as phonemic, etc. Let's just use the IPA-all template and write . Help:IPA explains that $⟨⟩$ can be used for, as does the Handbook of the IPA. Kbb2 (ex. Mr KEBAB) (talk) 08:18, 9 August 2018 (UTC)
 * I'm specifically talking about differences that can be explained using our diaphonemes. I'm not saying $\langler\rangle$ would be wrong in a transcription enclosed in brackets. I'm saying using IPA-all (or something like it) loses the information about the value of each sound readers could gain by clicking or hovering on the notation if IPAc-en were used. Nardog (talk) 08:39, 9 August 2018 (UTC)
 * I suppose you have a point. The acoustic distance between and  (or ) is huge and is much larger than, say, the distance between RP  and GA  in speakers that don't have the cot-caught merger. Kbb2 (ex. Mr KEBAB) (talk) 08:47, 9 August 2018 (UTC)
 * Also, what's the point of a notation of a local pronunciation that can be reproduced only by the locals? Nardog (talk) 08:51, 9 August 2018 (UTC)
 * Are you just thinking out loud or did you change your mind? Because I'm not sure what you mean. Kbb2 (ex. Mr KEBAB) (talk) 08:54, 9 August 2018 (UTC)
 * I'm making another point that I believe supports my position. Only those who do not have the foot–strut split will pronounce as [ˈbʊri]. Doesn't that kind of defeat the purpose of having a local pronunciation in the first place? Those who have the split will not register [ˈbʊri] as  upon hearing it, nor will they think they have to say [ˈbʊri] if they want to mimic the local pronunciation upon seeing "". Nardog (talk) 09:05, 9 August 2018 (UTC)
 * I agree, except for the very last point. I think that, at least in Britain, it's common knowledge that Northern English dialects lack the foot-strut split and that the quality of the merged vowel is generally more like than . Kbb2 (ex. Mr KEBAB) (talk) 09:29, 9 August 2018 (UTC)
 * My first and third impression (ha... I'm going back and forth on this a lot) is to agree with you, tentatively. So the real question is can we can up with any counterexamples? Wolfdog (talk) 12:50, 9 August 2018 (UTC)
 * Counterexamples to what? And who do you mean by "you"? Nardog (talk) 12:59, 9 August 2018 (UTC)
 * OK, sorry... you, Nardog. I'm saying we would have to think of counterexamples, meaning examples of articles/pronunciations where the diaphonemic system would not be the best-case scenario. Couldn't there be some feasible reason of interest for providing in rare cases a phonetic transcription (beyond the diaphonemic) of a name? Wolfdog (talk) 13:09, 9 August 2018 (UTC)
 * You mean cases like Kenya, Baltimore, Ska, and Baltimore, County Cork (whether these are appropriate or not)? I'm not talking about those. Again, I'm specifically talking about differences that can be explained using our diaphonemes. Whenever a transcription is intended to illustrate a subphonemic difference, the use of IPA-all is totally fine by me, as MOS:PRON recommends. Nardog (talk) 13:32, 9 August 2018 (UTC)
 * Using the diaphonemic system to indicate "local" pronunciations is fine for cases where the local pronunciation marks a difference in diaphonemic incidence (that is, they use one diaphoneme instead of another) and as long as we use IPAc-en, which is visually distinct (with its dotted underline and slashes) from those "local" pronunciations that use phonetic brackets and as long as we put actual diaphonemic transcriptions consistent with the guide.
 * It wouldn't make sense in cases where the "local" pronunciation uses the same diaphonemes but with different realizations. This includes when there are different phonemes in this "local" pronunciation because of a lack of phonemic contrast. Adding a "local" pronunciation for Northern English dialects that don't distinguish between and  would be redundant because NE readers are already tasked with reading  as  anyway. This principle is how we can avoid people putting a "local" pronunciation that is just a non-rhotic version of the diaphonemic transcription. Non-rhotic speakers are tasked with reading  as  and  as  already, so an additional transcript that just differs on the presence of r would be unnecessarily repetetive. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt]  16:21, 9 August 2018 (UTC)
 * So what would you do with Bury? The RP pronunciation has instead of, so maybe you could argue that the pronunciation with  is the underlying one (it's certainly older in all  words, for obvious reasons). Kbb2 (ex. Mr KEBAB) (talk) 22:24, 9 August 2018 (UTC)
 * This is where dictionaries can help us out. If we can't sufficiently glean a diaphonemic transcription from what sources give, then we'd want to avoid giving a diaphonemic transcription. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [lɛts b̥iː pʰəˈlaɪˀt] 04:45, 10 August 2018 (UTC)