Help talk:IPA/English

rhotic diaphonemes followed by vowels

How am I supposed to indicate these? You treat /ɔr/ as a (dia)phoneme, indicating either /ɔ˞/ or /ɔː/, depending on the rhoticity of the dialect. So does that mean I need an additional /r/ to indicate the linking-r found in non-rhotic accents?

My guess is no, based on the use of "Laura" as an example, but, either way, you really need to describe linking-r somewhere on this page, and give some example with two syllables in the hovertip in {{IPAc-en}}. (Unfortunately, my accent does not differentiate between /ɔr/ and /ɔːr/, so I don't know what common English word would be appropriate.) — trlkly 22:44, 5 August 2015 (UTC)

If you are talking about the linking r found among non-rhotic speakers in phrases like cheater eater, we indicate that with the /r/ after the /ɔ/. If you're talking about the intrusive r found in phrases like cheetah eater, we don't indicate that in the diaphonemic system.— Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 04:17, 6 August 2015 (UTC)

Opinion needed, and maybe correction?

I tried to add pronunciation to one article, Bill Doran (motorcyclist), the first attempt I had made and will probably be the only attempt, formatted as (/ˈdɒrʌrn/ DORR-UN). It was deleted within two days with the edit summary "not possible pron".

Could someone please check what I submitted [1] and show a corrected format here as necessary, so that I can amend the article? I guessed that the deleting editor was better-positioned than I, but I was disappointed that no correction was offered. The only thing I can see now is that the unstressed syllable is shown in upper case, and maybe should be DORR-un? I don't need to greatly understand what I've done wrongly, but I need to clear up this loose end. Thanks.-- Rocknrollmancer (talk) 21:10, 29 August 2015 (UTC)

If it rhymes with "foreign" and "sporran" then it should be /ˈdɒrən/. Rothorpe (talk) 21:19, 29 August 2015 (UTC)

Thanks, you've got it. How about the respell part? --Rocknrollmancer (talk) 21:22, 29 August 2015 (UTC)

Is the schwa symbol used there? If not, your suggestion DORR-un looks good. Rothorpe (talk) 21:41, 29 August 2015 (UTC)

We use schwas in respell too.

I am the one who deleted. The reason I didn't fix is that I had no idea if the o vowel was ɒ, ɔː, or oʊ. (That is, DORRən, DAWRən or DOHRən.) — kwami (talk) 21:49, 29 August 2015 (UTC)

OK, much obliged to you both. Remember I am a beginner and this was nine months ago - I need to re-add it, and to another similar article Doran's Bend. I believe the schwa should be included. For respell, Dor as in moribund, an as in run. Formatted as {{respell|DORR|{{sc|un}}}}, the template will only show the second syllable as upper case DORR-UN --Rocknrollmancer (talk) 22:06, 29 August 2015 (UTC)

Should be a schwa. I seriously doubt it's like "run", unless he pronounces his name as a compounded "Dor-Un". And I don't understand why you're adding in the {{sc}} template. — kwami (talk) 23:24, 29 August 2015 (UTC)

Yes, we've established that it's a schwa in IPA, so it must be in the respelling too. Rothorpe (talk) 23:40, 29 August 2015 (UTC)

OK, thank you. I'm using a 1960s interview given to the journalist by his daughter, and I've not looked at it since December 2014, so it's always going to be a matter of interpretation from the source. I don't know why the {{sc}} is there, either; it's something I worked out at the time, seemingly wrongly. The TV commentator used the exactly-correct pronunciation, IMO and interpretation of the 1960s article, recently on UK television's ITV coverage. I can sort out the exact quote and put it here, Sunday UK daytime, if that might help. I think you're not far off with Dor-un, but with the emphasis on a slightly-long rr. If I don't do it, the chances are it will never get added to the articles, and it would be assumed to be Dor-Ann.--Rocknrollmancer (talk) 00:02, 30 August 2015 (UTC)

Well, you've already OK'd the IPA, so I've put that in. Rothorpe (talk) 00:46, 30 August 2015 (UTC)

Thanks very much. Regrettably, I was unable to trace the citation I used which is embarassing - turns out to be 1973 newspaper-format, not a 1967s magazine as I thought...maybe I have a scan somewhere on the 14-year-old computer that is groaning with research stuff...Getting back to basics, the example of 'sporran' would be most appropriate. I had a conversation with my brother (who lives in US) some years ago regarding their pronunciation of 'Fillet' (fish, meat) as Fee-lay, or sometimes contracted to F'lay; I know Wikipedia deprecates the use of apostrophe, but in this vein I would write Doran as Dorr'n. rgds.--Rocknrollmancer (talk) 12:37, 4 September 2015 (UTC)

BBC/RP pronunciation

The article states: "even amongst educated speakers. BBC English is moving away from the older RP [æ] towards the more open vowel [a]".

I think this is seriously out of date.

At first I couldn't make any sense of it, but then I remembered hearing some very old (50's/60's) public information films where the narrator spoke in such a way that 'lad' could conceivably rhyme with 'lard'. To give an idea of how old this is, people start laughing as soon as the narrator speaks - the accent is considered hilarious.

Nonetheless, I'd like to hear if there are other opinions before changing the tense of that fragment, particularly as it's hard to back up with references. PRL42 (talk) 09:57, 4 September 2015 (UTC)

'lad' rhyming with 'lard'? That's never been a part of RP. Peter238 (talk) 11:16, 4 September 2015 (UTC)

No, it would be more someone doing a verbal caricature of an absurdly 'hoity-toity', old fashioned, upper class accent. I see you've now removed the mention of the word. PRL42 (talk) 13:55, 4 September 2015 (UTC)

Source? You must be confusing vowels. In the old-fashioned upper-class accent of UK (which is a form of RP, called Upper Crust RP or Conspicuous General British), /æ/ had a [æ ~ ɛə̯] quality, often with some pharyngealization, so that bad and bared had a quite similar vowel (the latter tended to be longer and lack the pharyngeal quality of /æ/). /ɑː/, on the other hand, was back [ɑː] or central [äː] (but not front, unlike /æ/ which is always front in RP), without the pharyngealization, and had a similar quality to the vowel in the word nurse (which was [ɐː]). The [a] quality of /æ/ is a fairly recent innovation, first described in 70's/80's (e.g. in Wells's "Accents of English", which is from 1982). Peter238 (talk) 15:20, 4 September 2015 (UTC)

PRL42, perhaps you're confusing [a] with [ɑː]? The two are different. In technical phonetic usage, [a] is front or central, [ɑː] is back. The point about lad being pronounced as [lad] is that it's open rather than near-open [æ] or open-mid [ɛ], not that it's back open like lard [lɑːd]. — Eru·tuon 15:36, 4 September 2015 (UTC)

Jump links to etymology and pronunciation

Where will Wikipedia go on this topic brought up a few months ago? A discussion of it is underway at Wikipedia talk:Manual of Style/Medicine-related articles#Simple first sentence. Quercus solaris (talk) 23:22, 4 September 2015 (UTC)

The first link doesn't work. Peter238 (talk) 07:04, 5 September 2015 (UTC)

It was just archived. I've fixed the link. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:30, 5 September 2015 (UTC)

Variable vowel a:

I am quite uneasy about this innovation. I haven't seen it used elsewhere, and I think it's unlikely many WP users will understand how it works. They would have to grasp that it's OK to transcribe 'bath' as /ba:θ/ but not OK to transcribe 'cat' as /ka:t/. Can anyone give a reference for where this convention is used? RoachPeter (talk) 10:40, 14 September 2015 (UTC)

A few of our variable vowels are technically original, though AFAIK the variable pronunciations are represented in dictionaries. If variable a represents BATH, then things are still fairly clear (at least as far as finding sources to back up our transcriptions). I'm not so certain about the way it's been extended, though. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 20:26, 14 September 2015 (UTC)

The footnote about ɵ gives a self-referencing pronunciation

/ɵ/ is pronounced as [ə] in many dialects, and [ɵw] or [əw] before another vowel, as in cooperate. --V111P (talk) 00:17, 22 September 2015 (UTC)

/ʊr/

Apart from courier, is the sequence /ʊr/ found in any other words? 200.77.119.71 (talk) 06:51, 5 October 2015 (UTC)

Wiktionary has a list here. There are also a number of proper nouns, such as Couric, that use this sequence. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 08:07, 5 October 2015 (UTC)

Only rarely in RP. There are quite a few words that might contain this sequence but only (according to the dictionaries) as an alternative, e.g. 'natural' is usually /'nætʃ.ə.rəl/ but could be /'nætʃ.ʊ.rəl/ (giving an /ʊr/ sequence). But cases of /ʊr/ within a stressed syllable, as in 'courier', are very rare and are mostly in recently imported words or names (e.g. Fourier, guru). Other words containing /ʊr/ but not in a stressed syllable are tureen, purpura (second syllable), Touraine, Jurassic, kurus. RoachPeter (talk) 08:26, 5 October 2015 (UTC)

/i/

The page currently says that the variable vowel /i/ represents either /ɪ/ or /i/. This is an error, because /i/ without the length symbol isn't listed elsewhere in the table. It isn't listed because there are only two close front vowels, /iː ɪ/; /i/ isn't a third close front vowel phoneme, it's just an unstressed vowel that can be one of the two depending on dialect, though in most standard pronunciations today it's really just /iː/. See Geoff Lindsey's post on this.

I changed the text to reflect this, but Dbfirs reverted me, saying that the OED uses /i/ in serious. Actually, the version of the OED that I can access through my library says /ˈsɪərɪəs/, but yeah, the OED does use the symbol /i/ in city /ˈsɪti/. Still, this doesn't mean there's actually a phoneme /i/ separate from /iː/; the use of a separate symbol is just a nod to the fact that more old-fashioned speakers said /ɪ/. The last vowel in city is identical to /iː/, as it's often diphthongized to /ɪj/. At least, this is what Lindsey says and he makes a convincing case. — Eru·tuon 17:56, 9 October 2015 (UTC)

I suspect that there is some equivocation going on between phonetics and phonology here. As far as phonetics, it is my understanding that the final [i] of city is both short and tense, which makes it different from a long tense vowel or a short lax vowel. This pronunciation contrasts dialectally with those who pronounce this vowel as [ɪ]. So, phonetically speaking, there is variation between [ɪ] and [i].

However, this is a diaphonemic transcription and we are largely bypassing phonetics. As such, the issue is more about whether this final vowel can be grouped with the allophones of /ɪ/ or the allophones of /iː/. For those who pronounce this vowel as [ɪ], the issue is clear. For those who pronounce it as [i], it's ambiguous. This seems to be the way we want to present the issue to a lay audience. So it makes sense to keep the length mark in. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 20:25, 9 October 2015 (UTC)

I was looking at the big OED. To me, the use of /iː/ is unreferenced and misleading (given the table in the article), but I agree that there is a lot of disagreement over the interpretation of this. In what version of the language is "The last vowel in city ... identical to /iː/"? I've never heard it (except in Michael Crawford's Bettee), and I can't find any dictionary that uses the long version. Dbfirs 08:12, 10 October 2015 (UTC)

Ƶ§œš¹: Actually, the final vowel of happy isn't necessarily short. As you can hear in the soundfiles in Geoff Lindsey's post, speakers of "Standard Southern British" (Lindsey's term) often diphthongize it, so that happy is [ˈhapɪj] (or as I wish Lindsey would write it, [ˈhapej]), not [ˈhapi]. The same is true in Australia, where it is often pronounced with a centralized onset: ˈhæpəi]. (This comes from personal observation, not from Lindsey, since I've been watching an Australian show.) American English might have a short and tense vowel, but probably just because there's less diphthongization of fleece and vowel length isn't usually considered as phonemic. But in SSB (modern RP) and AuE, the vowel is most often diphthongal, not short.

Dbfirs: I'm not that familiar with the transcriptional practices of dictionaries, so I can't provide any counterexamples to your point about the OED. However, please read the article by Lindsey above, to see why the OED's practice is misleading. The final vowel of happy isn't actually different from fleece in dialects with happy-tensing, despite the way the OED implies it is. — Eru·tuon 01:32, 22 October 2015 (UTC)

This is the sort of problem that is caused by the in-house tradition of using diaphonemes. They are a relatively obscure concept that is not used in many analyses. However, they are (implicitly?) used in some major dictionaries. --mach 🙈🙉🙊 01:46, 22 October 2015 (UTC)

It's a relatively minor problem compared to the alternative(s). — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 04:44, 23 October 2015 (UTC)

I feel like the discussion is going way off topic, onto what the vowel of happy sounds like. I'm a little out of my depth in trying to prove that HAPPY is like FLEECE. I'm not a professional phonetician or phonologist, and in all my wordiness it seems like nobody's getting the point.

I'll try another line of argument; maybe I'll make more sense this time.

The table currently says the variable vowel /i/ is equal to /ɪ/ or /i/. /ɪ/ occurs elsewhere in the table, but /i/ only occurs in the variable vowels section.

So, the variable vowel is equal either to a regular vowel or to itself. x is equal either to y or to x. Then what is x? x is never defined. The statement is tautological and therefore meaningless.

Similarly, the variable vowel /i/ is never defined, since it's said to be equal to itself. What the table is supposed to say is that /i/ represents either /ɪ/ or /iː/. This would be meaningful, because /ɪ/ or /iː/ occur outside of the variable vowels section. — Eru·tuon 05:35, 22 October 2015 (UTC)

I sort of see what you are getting at, but there is so much regional variation in the actual pronunciation of some vowels that I don't think we will ever agree. You seem to be convinced by Geoff Lindsey's opinion. I'm not. Here in northern England, I don't even hear what he seems to be hearing in his sound extracts, especially for Kasia Madera's party, and I think he has been very selective in his choice of lengthened vowels from Prince William's speech (though I would regard William's accent as slightly "off" from standard BBC English anyway). Personally, I prefer to stick with the OED where /iː/ is used only for truly long vowels. Dbfirs 07:29, 22 October 2015 (UTC)

A compromise could be “(either /ɪ/ or /i/ or /iː/)”. I would keep /i/, since there are phonemic analyses of English that use /i/ precisely for this reduced vowel. It is not uncommon that a phonemic analysis may assume different sets of vowels for stressed syllables and for reduced syllables (think /ə/). At any rate, it is important that we use the slashes for indicating phonemes. --mach 🙈🙉🙊 08:32, 22 October 2015 (UTC)

I don't like that as a compromise. It would be like saying the second consonant in butter is /t/ or /ɾ/ or /d/. Alveolar flapping is a notable feature of a number of dialects, but there is no separate alveolar flap phoneme. From what I gather, there is no linguistic analysis (or, at least, none in major dictionaries) that would regard this vowel as a separate phoneme. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]}

According to John Wells’s overview of IPA transcription systems for English, there are several major dictionaries that use /i/ even though he only looks at British English dictionaries. --mach 🙈🙉🙊 15:22, 22 October 2015 (UTC)

J. 'mach' wust: Your compromise still doesn't make sense. You're saying we should say x is equal to y, x, and z. Now, y and z are defined on this help page, but x is defined in the OED and other dictionaries. That doesn't make sense. This help page is supposed to be a self-contained logical system. It has to refer to itself; it can't refer to symbols defined in other transcriptional systems. If we refer to other transcriptional systems, why don't we note that the symbol /æ/ is equivalent to the symbol /a/ used in the current online edition of the OED, that /ɛər/ is equivalent to /ɛː/, and so on? We don't, because this page is supposed to be a self-contained transcriptional system, not a list of all the alternative transcriptions used in other dictionaries. — Eru·tuon 20:43, 22 October 2015 (UTC)

I thought a bit, and there's one thing that could be done to resolve the logical inconsistency: moving /i/ out of the variable vowels section and into the unstressed vowels section. That would express the fact that several people here think it's a separate phoneme, not just a disjunction between two other phonemes, and it would remove the ridiculousness of saying that a variable stands for itself. What do you all think? — Eru·tuon 21:39, 22 October 2015 (UTC)

It is not a variable, but a variable vowel. I see no obstacle in it standing for itself. But I agree that moving it to the reduced vowels is a good solution. After all, it cannot ever be stressed. Here follows the answer I was about to post when your own reply came first:

Most symbols on this help page are defined by referring to themselves. Where it says: “ʌ | STRUT, bud, dull, gun”, you might as well read: “ʌ STRUT, bud, dull, gun (always /ʌ/)”. This would be analogous to saying: “i | HAPPY, serious (always /i/)”. However, the symbol “i” can not only stand for itself (like most other symbols), but it can also stand other symbols (like the other variable or reduced vowels). It is variable, and one of the variants it may stand for is itself: “i | HAPPY, serious (sometimes really /i/, but alternatively also /ɪ/ or /iː/)”.

The entire thing is an overly complicated mess because everything is forced into the Procrustean bed of a diaphonemic analysis that includes several violations against the IPA usage and has been originally created on the English Wikipedia (I fear it is diffusing into the internet). --mach 🙈🙉🙊 21:50, 22 October 2015 (UTC)

I also agree with moving it to the reduced vowels section. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 04:42, 23 October 2015 (UTC)

 Done — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 05:09, 23 October 2015 (UTC)

I also agree with that solution. Dbfirs 07:18, 23 October 2015 (UTC)

J. 'mach' wust: I defended the Wikipedia diaphonemic system in a previous discussion, but I'm beginning to see your point of view. I recently responded to someone (Peter Greenwell) who was confused because the vowel of Australian English bed is listed in the article on the Close-mid front unrounded vowel. He thought, because the Wikipedia diaphonemic system transcribes the vowel of bade as /eɪ/, that the vowel article is saying that Australian bed [bed] rhymes with bade, which would be transcribed as /beɪd/ under the Wikipedia diaphonemic system. But phonetically Australian English bade has a different vowel sound, [æe], so bed [bed] and bade [bæed] do not rhyme. The Wikipedia transcription is based on a now very old-fashioned–sounding posh British pronunciation, and doesn't represent the pronunciation in AuE.

His confusion is a direct result of the Wikipedia diaphonemic system, which he was interpreting as literal phonetic fact, rather than as diaphonemic abstraction. He's only one person, but if there are others like him, the system is causing problems, as W. P. Uzer suggested it would in the discussion earlier this year (now to be found in Archive 13). I stand by my statement that we needed evidence of reader confusion, but now some evidence is coming in. Anyway, I've gone off on a tangent quite unrelated to the topic of this thread. Sorry. — Eru·tuon 07:47, 23 October 2015 (UTC)

Yes, I think many people are confused by any diaphonemic system when they first meet it (I know I was, and occasionally still am). Any such system can only be an approximation, and each of us applies our own "corrections" to match our own pronunciation; for example, I always read /æ/ as /a/ for modern British English. Dbfirs 08:16, 23 October 2015 (UTC)

I got a ping from this discussion, so will just quickly add my usual 2p - we ought to use standard notations in standard ways, preferably not invent non-standard ones, but certainly not use standard ones in non-standard ways. Use of /i/ for the happy vowel is standard as far as I'm concerned (part of the IPA-based system used by major dictionaries). It's things like /a:/ for the BATH diaphoneme, a misleading/incomprehensible Wikipedia invention, that need to be eliminated. W. P. Uzer (talk) 08:29, 23 October 2015 (UTC)

Yes, that is misleading because I don't know of any region that pronounces it [baːθ]. In northern England and Wales it's [baθ], of course, and [bɑːθ] in southern England, and on the other side of the pond it's often a variant of [bæθ], but /baːθ/ seems to be a Wikipedia invention. Dbfirs 13:21, 23 October 2015 (UTC)

/aː/ is not meant to be a phonetically accurate symbol, but a symbol that means "either /æ/ or /ɑː/" (phonetically (in case of the whole UK and Ireland), /æ/ is also not very accurate anymore). We couldn't use /ɑː/ for that purpose, because it is already used to transcribe, well, /ɑː/. That said, /ɑː/ is central [äː] or front [aː] in Australia and New Zealand, and supposedly also in Norfolk (only the central variant). Peter238 (talk) 13:32, 23 October 2015 (UTC)

Oh, yes — apologies to our antipodean friends (and Bernard Matthews?) for forgetting their pronunciation. I suppose the symbol looks correct to them. I realise it's not supposed to be phonetically accurate, so I'll just get used to shortening it to [a] for my local dialect, just as I read /æ/ as [a]. (Your conversions will be different.) Dbfirs 14:22, 23 October 2015 (UTC)

The underlying problem is that it is NOT a symbol that means that at all. Where did Wikipedia linguists get this strange idea that they can just appropriate standard IPA symbols for their own ad hoc purposes? The intended meaning will only be understood by someone who clicks through to this page and somehow deciphers its content. And since IPA is allegedly familiar to people already (at least, that's the rationale for our using it), there's absolutely no reason why anyone would click through to this page. The situation is absurd, as I've pointed out countless times, but for some reason it still appears not to have been changed, there are just tweaks made to the page that many users are not going to look at in any case. W. P. Uzer (talk) 16:06, 23 October 2015 (UTC)

All I did is described how this page and transcriptions linking to it use the /aː/ symbol. You don't have to list all of the problems this transcription might create (or does create), because I'm aware of them. Trust me, I read a large portion of the countless discussions you're referring to. Peter238 (talk) 16:15, 23 October 2015 (UTC)

Not addressed to you specifically; others might not be aware of them. If enough people become aware, something might finally be done about it. W. P. Uzer (talk) 16:26, 23 October 2015 (UTC)

Oh, ok. Peter238 (talk) 16:35, 23 October 2015 (UTC)

Actually I've just looked and I've seen something even more ridiculous has happened - the /a:/ symbol now no longer means RP /æ/ GAm /ɑː/, it now apparently means either that OR THE REVERSE. Like, what?? How is anyone now supposed to derive any information from that symbol at all, even if they do happen to have chanced on the page that explains it? W. P. Uzer (talk) 16:10, 23 October 2015 (UTC)

Yep, that appears to be a problem. Maybe let's reserve /a:/ for just the words with the bath-trap split, nothing else? As far as I can remember, that's what I proposed when /a:/ was introduced. Peter238 (talk) 16:15, 23 October 2015 (UTC)

Yes, better, but in fact /a:/ should never have been introduced at all. Can you not see that no-one will possibly know what it means until it's explained to them (and that most users will not be seeking an explanation, since it looks like a standard IPA symbol)? And what is the possible need for such a notation anyway? W. P. Uzer (talk) 16:24, 23 October 2015 (UTC)

There's no real need for them. The only reason /aː, ɒː/ are used on this page is to have one transcription instead of two. Which may be really confusing in non-BATH words. However, /i, u/ should stay, as e.g. Longman Pronunciation Dictionary uses them in all of its transcriptions (albeit only in unstressed positions). Peter238 (talk) 16:35, 23 October 2015 (UTC)

@W. P. Uzer: I fully agree with you that we should never have invented our own conventions. I think it is really a violation of WP:OR. --mach 🙈🙉🙊 21:16, 23 October 2015 (UTC)

Removing /a:/ and /ɒ(:)/

From past discussions and this one, I believe there is general agreement, albeit sometimes reluctant, that at the very least this /a:/ and /ɒ(:)/ convention is wrong on several different levels (OR, misleading, etc.) We should eliminate them from the system, I think, and agree to use multiple transcriptions for such words - there is really no other way to effectively convey that information to readers (and remember we are not only conveying information about pronunciation, but to some people we are conveying - currently wrong - information about the IPA and normal use of it). Another question the problem of where pronunciation differs due to non-rhoticity - others have disagreed, but it seems very wrong to me to give only General American (or mid-Atlantic) pronunciations of place names in Britain and Australia, for example. I don't have any problem with the reduced vowels, though, since as Peter says, we are following what standard sources do (though if we are choosing to use slightly different symbols than they do due to rendering issues, it would be good to check that those issues really exist for large numbers of people). W. P. Uzer (talk) 07:05, 25 October 2015 (UTC)

At least with ⟨ɵ⟩, we are not following any standard source. It is pure WP:Original research. And so is the use of the symbols ⟨ɨ⟩ and ⟨ʉ⟩ instead of the OED’s ⟨ɪ⟩ and ⟨ʊ⟩. The use of ⟨ɪ⟩ ⟨ʊ⟩ is as unproblematic as the use of ⟨ɪ ʊ⟩. If anything, there might be display issues when using ⟨ᵻ ᵿ⟩. Other WP:Original research uses include the ⟨u⟩ for “variable” /ʊ/ or /uː/ or distinctive ⟨.⟩, maybe also ⟨oər⟩ for /oʊr/ and ⟨ɔɪər⟩ for /ɔɪr/.

It might be that for the first time since the creation of this help page, there is a window of opportunity for reverting its blatant violation of WP:OR that was criticized from the start. The user who most fervently advocated for this OR appears not to participate in this discussion.

I think we should prepare a reasonable proposal. Research what’s really being used – outside of Wikipedia so we do not end up in the WP:OR trap, and inside of Wikipedia (I suspect the controversial signs are not really used except by fervent advocates of this help page).

For things such as the rhotic/non-rhotic differences, the quarrel about dialect differences is an irrelevant straw man. The relevant question is: What is WP:OR and what is not? Are there dictionaries from non-rhotic regions that indicate the r’s and depend on the fine print for dropping them (as we do currently)? Are there dictionaries from rhotic regions that do not indicate the r’s and depend on the fine print for inserting them (such a solution would be equally viable)? Are there compromises?

File:Verifiability and Neutral point of view (Common Craft)-en.ogv

Short video explaining the concepts of "Verifiability", and "Neutral point of view"

When we are prepared, we should get approval, maybe at the Wikipedia:No original research/Noticeboard. We should stress that this is really about the WP:Core content policies and that it really concerns the main content of the articles, not just a help page. Then we can make our move and finally cast out the OR. --mach 🙈🙉🙊 08:14, 25 October 2015 (UTC)

We might also consider how many articles would be affected by removing /a:/ and/or /ɒ(:)/ or by restricting the former to just BATH vowels. Is there a way to check that? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 17:56, 25 October 2015 (UTC)

Something to consider with this proposal is that the key difference between our usage of /ɔː/ and /aː/ is that, for the former, those who do what we ask and ignore the distinction between /ɔː/ and /ɑː/ if they have merged the two vowels typically ignore this distinction in real life; cot-caught merging speakers like myself typically perceive no difference between the vowels. On the other hand, the distinction between /æ/ and /ɑː/ is marked for all speakers. With the former, we're asking readers to do what they normally do, while the latter we're asking them to do something new. This newness may be negatively impacting the readability of our system. I can't speak to /ɒː/ in this regard, though, it may be similar to /aː/. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 20:02, 1 November 2015 (UTC)

I agree. Introducing the ⟨aː⟩ and ⟨ɒː⟩ symbols is novel and not backed by WP:SOURCES – which means it is WP:Original research. Also, it is potentially misleading because ⟨aː⟩ could easily be mistaken for /ɑː/ – there are indeed transcription system that use ⟨aː⟩ in that way. Using ⟨aː⟩ and ⟨ɒː⟩ is almost as ridiculous as using a hypothetical novel *⟨ɑeɪ⟩ symbol so we could transcribe two pronunciations of tomato in the same way as */təˈmɑeɪtoʊ/ (please never ever consider that seriously). ☺ --mach 🙈🙉🙊 21:39, 1 November 2015 (UTC)

You may laugh, but {{IPAc-en}} currently encodes for a number of superscript letters that have the same logic as this and aren't even mentioned. I've called for their removal, though I don't think anyone has done anything about it. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 22:27, 1 November 2015 (UTC)

There is a workaround for finding the articles that contain the contested signs. Unfortunately, you cannot use wildcard search terms like *aː* because using a wildcard character at the beginning of a search term does not work. However, it works perfectly well when the wildcard is prepended with another letter, e.g. a*aː*. This has to be repeated with every letter of the English phonetic alphabet (including signs such as ⟨ˈ⟩). It is laborious, but doable. By including an additional linksto:"Help:IPA for English", the results can be narrowed down considerably. This gives the following search terms:

• a*aː* linksto:"Help:IPA for English"
• æ*aː* linksto:"Help:IPA for English"
• ɑ*aː* linksto:"Help:IPA for English"
• b*aː* linksto:"Help:IPA for English"
• d*aː* linksto:"Help:IPA for English"
• etc.

It is laborious, but doable. --mach 🙈🙉🙊 17:01, 11 November 2015 (UTC)

What's wrong with just putting 'aː* linksto:"Help:IPA for English"'? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:26, 11 November 2015 (UTC)

Sure, aː* linksto:"Help:IPA for English" has to be included as well – but it only matches words that begin with aː. --mach 🙈🙉🙊 19:49, 11 November 2015 (UTC)

Note on variable vowels

@J. 'mach' wust:, I like your idea of alerting readers familiar with the IPA that we use some IPA symbols in unconventional/novel ways. I've modified it so that it's one note, rather than three separate ones. I think we could also include mention of our other variable vowel symbols in that note. I'm not wedded to my particular wording, though "not used according to the IPA" did rub me the wrong way a bit. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 05:09, 23 October 2015 (UTC)

I understand that "not used according to the IPA" does not sound very nice. But that is just what it is: The symbols are not used for the sounds they represent according to the IPA. I think honesty to the readers demands that we notify them of this fact. It is unusual and thus highly noteworthy. Consider that the symbol “r” also gets a note about an unusual usage, even though it is used in major dictionaries.

And I do not quote understand why you have moved the symbols “̵̵ɨ, ʉ, ɵ” to the variable vowels. Of course, they are variable. But so is “i”. I would rather keep all the reduced vowels together, and use the variable vowels section only for those that can receive stress. --mach 🙈🙉🙊 06:19, 23 October 2015 (UTC)

I think the problem I have with "not used according to the IPA" is that it sounds like we are the only ones who deviate from the strict phonetic usage of, e.g. [ɨ] as a close central unrounded vowel. We aren't. What is notable about our in-house convention is that we are using one symbol to represent a difference in phonemic incidence, rather than a separate phoneme.

I also disagree with your splitting this information into three separate notes. There's already an explanatory note for each of those three symbols and what they mean. Having a separate note for each one seems excessive and unnecessarily makes the information more difficult to access for the reader. I won't revert your (partial) revert at this point; let's come up with the wording we all can agree on. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:47, 23 October 2015 (UTC)

Incidentally, since this use of ɨ is apparently based on the latest practice of the OED, why do we not choose to use the exact same symbol that they do? W. P. Uzer (talk) 16:21, 23 October 2015 (UTC)

Rendering issues I think. Peter238 (talk) 16:31, 23 October 2015 (UTC)

That is correct, rendering issues on Ƶ§œš¹’s Android phone, see Help talk:IPA for English/Archive 14#Sources for non-IPA use of the symbols ⟨ɨ⟩, ⟨ʉ⟩ and ⟨ɵ⟩.

@Ƶ§œš¹: I have of course no objection against having a single note for the three signs. That is why I have kept a single note like you did. I just think the links to the note are more appropriately placed with the individual signs: One note – three links.

I do not think that "not used according to the IPA" sounds like we are the only ones who deviate from the IPA. To me, it just says that the signs are not used according to the IPA. I think the deviation from the IPA is equally notable as the use of diaphonemes and should be made explicit. Not declaring the deviation from the IPA looks to me as if we were trying to hide it, hoping we get by without anybody noticing our deviation.

In our in-house convention, a sign like our ⟨ɨ⟩ can stand for several phonemes – /ɪ/, /i/, or /iː/ – but it does not stand for IPA /ɨ/. The sound of IPA [ɨ] apparently exists in some Southern English pronunciations of /uː/, but these would not even be spelt with ⟨ɨ⟩ in our in-house conventions. I think our ⟨ɨ⟩ is more problematic than our ⟨aː⟩ for the vowel of words like bath. At least the individual components ⟨a⟩ and ⟨ː⟩ are really being used in certain transcriptions.

It looks as if our ⟨ɨ⟩ were chosen because of its graphical similarity to ⟨i⟩, just like the OED ⟨ᵻ⟩ (or ⟨ɪ⟩) looks as if it were chosen because of its graphical similarity to ⟨ɪ⟩. The OED ⟨ᵻ⟩ is less misleading, though, because it does not belong to the IPA in the first place. A person who knows about the IPA will notice that the strange thing about a transcription such as /ˌæpəˈlætʃᵻn/ is the non-IPA sign. When they see a transcription such as /ˌæpəˈlætʃɨn/ (as in the article about the Appalachian Mountains), they might assume that the strange thing about it were an obscure Appalachian dialect. --mach 🙈🙉🙊 21:05, 23 October 2015 (UTC)

I see that I mistook your edit as a revert to your three-note version, my apologies.

It's apparent that you see the situation in relation to IPA usage in a different manner than I do. As you are okay with not using the phrase "not used according to the IPA" in regards to the explanatory note for the English rhotic, presumably because this is standard practice across scholarship and is therefore not simply an in-house practice, you are equivocating two separate things: phonetic accuracy of our symbols and the use of symbols to represent more than one phoneme across dialects. Since we are striving for clarity, even for those familiar with the IPA, we want to focus on the latter, since this is the more unusual thing about our system.

In this system, we use ⟨ɔː⟩ and ⟨ɑː⟩ to represent a vowel contrast that not everyone makes. To oversimplify the matter, some people contrast [ɔː] and [ɑː] (with a minimal pair being cot and caught) and others pronounce both as [ɑː]. In a similarly oversimplistic manner, some people (including myself) contrast a lowered [ɨ] with [ə] (with a minimal pair being roses and Rosa's). This is discussed in article space here. The former vowel is often analyzed as an allophone of /ɪ/, and our focus here is on grouping phonemes, but the central pronunciation is certainly there and may actually be the justification for using the capital barred I symbol in the OED.

Your preference for the phrase "not used according to the IPA" for /ɨ/ but not for /ɔː/ seems to be due to a misperception that people don't have a central pronounciation for the vowel we transcribe as ⟨ɨ⟩. If I'm incorrect in my assessment of your understanding, then I'm at a loss as to what justifies your disparate approach. IMHO, because there is phonetic justification for the use of the symbol, this would be more closely in accordance with the IPA, which technically eschews using ⟨r⟩ for the English rhotic.

If it's about phonetic imprecision, then it shouldn't be an issue, as ⟨ɨ⟩ is the most accurate IPA symbol we can use. If it's about representing two different sounds, no note is even needed as the whole system does that. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:53, 24 October 2015 (UTC)

What is the “phonetic justification” for the use of the symobl ⟨ɨ⟩, then? I am speaking WP:RELIABLE sources. Earlier this year, in the discussion Help talk:IPA for English/Archive 14#Sources for non-IPA use of the symbols ⟨ɨ⟩, ⟨ʉ⟩ and ⟨ɵ⟩, nobody has produced any sources. To my knowledge, the in-house use of ⟨ɨ⟩, ⟨ʉ⟩, and ⟨ɵ⟩ is

1. not according to the IPA (because there is no phonetic justification) and
2. a violation of WP:NOR (like ⟨aː⟩ or ⟨ɒː⟩).

--mach 🙈🙉🙊 17:15, 24 October 2015 (UTC)

I just gave you the phonetic justification for ⟨ɨ⟩ — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 20:43, 24 October 2015 (UTC)

Sorry, I had overlooked your “here” link that leads to an article with a source. I have no objection when you replace my "not used according to the IPA" by a proper citation. The use of ⟨ɨ⟩ in a diaphonemic transcription continues to be a violation of WP:NOR. And ⟨ʉ⟩ and ⟨ɵ⟩ continue not to be used according to the IPA. --mach 🙈🙉🙊 23:16, 24 October 2015 (UTC)

We've already discussed this matter ad nauseum and I've explained in detail how the information we present is grounded in sourcing, even with a unique and/or unconventional in-house system of presenting this information. I have also explained, in the very conversation you cite, how a strict application of NOR undermines our goals in a way that necessitates an WP:IAR approach. Remember, we're trying to be practical here. Our desire for a practical solution hasn't changed and you are still bringing up faults that are more theoretical than practical concerns. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 06:21, 25 October 2015 (UTC)

Just for the record: You had agreed that you ⟨ɪ, ʊ⟩ were OK to you if it were not for your Android phone.[2] My big mistake was that I forgot your Android phone has no trouble with displaying ⟨ɪ, ʊ⟩ so the matter died off once more. Nobody has produced any source for ⟨ɵ⟩ ever being used in phonemic transcriptions. --mach 🙈🙉🙊 21:20, 1 November 2015 (UTC)

I shouldn't be used as a spokesperson for technical issues. I was just saying what works on my devices. ⟨ᵻ, ᵿ⟩ seem like perfectly fine choices if they work. ⟨ɪ, ʊ⟩ are a kind of awkward to type, so between the two pairs I'd prefer the former. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 23:01, 1 November 2015 (UTC)

I would think that ⟨ᵻ, ᵿ⟩ are more awkward to type than ⟨ɪ, ʊ⟩ because it seems unlikely that the former exist on any keyboard layout. The latter may at least be typed with IPA keyboard layouts. Plus your Android phone is worth considering, after all. I admit I would personally prefer ⟨ᵻ, ᵿ⟩ because they are a cleaner Unicode solution, but I think that ⟨ɪ, ʊ⟩ are more practical. I also think there is no reason for getting all prescriptivist over this simple matter of styling. --mach 🙈🙉🙊 00:04, 2 November 2015 (UTC)

I didn't know there were IPA keyboard layouts. All of my IPA contributions have been clicking the links at the bottom or copying and pasting from a reference document on my computer. What do you mean by prescriptivist over styling? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 00:40, 2 November 2015 (UTC)

I guess the most popular IPA keyboard layouts are the SIL’s IPA Unicode Keyboards. I see they indeed contain the sign ⟨ᵻ⟩, so I was mistaken about that. My point about prescriptivism is that we should not care whether people prefer the styling ⟨ɪ⟩ or the Unicode character ⟨ᵻ⟩. Both are fine choices, even though the former is slightly more accessible (on devices like your Android phone). --mach 🙈🙉🙊 08:16, 2 November 2015 (UTC)

Our choices were not based on what was easy for us as wikipedians to type (after all, we can just type "i-" in the IPAc-en template and it will display whichever symbol we like), but rather on what the reader was more likely to have font support for. ⟨ᵻ, ᵿ⟩ are (or at least were) not universally available, while html strikethrough is inadequate because it is not copy&paste-friendly. (Note that all of our symbols copy true. That was a design feature.) We settled on ⟨ɨ⟩ because it suffered from neither of those problems. That is, the choice was purely practical.

⟨ᵻ, ᵿ⟩ are the letters I advocated until other editors convinced me they would not work well. If we can now use ⟨ᵻ, ᵿ⟩ without readers with old OS's and browsers seeing boxes, I'd be happy to switch over. — kwami (talk) 03:06, 3 November 2015 (UTC)

Whatever the original reasons were, now the material has been WP:CHALLENGED. This means it can be removed unless somebody provides relevant WP:SOURCES. --mach 🙈🙉🙊 09:07, 3 November 2015 (UTC)

Transcriptions that are probably WP:Original research

This is a list of transcriptions that are probably WP:Original research. Now we could put Template:OR in all the articles where such transcriptions are used. However, for sake of usability, I am putting Template:Original research on this help page. I know this help page is not an article in the strict sense. However, unlike other help pages its recommended transcriptions will be put as information into the articles.

Transcriptions that are probably WP:Original research include the following:

• The use of //aː// for /æ/ (in some notations, /a/) or /ɑː/
• The use of //ɒ:// for /ɒ/ or /ɔː/
• The use of //.// as a (dia)phonemic sign
• The use of //ɨ// for /i/ or /ə/
• The use of //ʉ// for /u/ or /ə/
• The use of //ɵ// for /oʊ/ or /ə/
• The use of //u// for /uː/ or /ʊ/
• The use of //oər// for /oʊr/
• The use of //ɔɪər// for /ɔɪr/
• The generalization of rhotic transcriptions with an explanation of non-rhotic pronunciation in the fine print instead of the generalization of non-rhotic transcriptions with an explanation of rhotic pronunciation in the fine print

We need to improve the help page by verifying the transcriptions and adding inline citations. Statements consisting only of original research should be removed. --mach 🙈🙉🙊 08:06, 27 October 2015 (UTC)

I don't understand the point. If a pronunciation is OR, it doesn't matter which sounds are in it. Or do you mean you don't like the transcription? That's irrelevant, since symbols are arbitrary.

Also, /./ is a syllable boundary. — kwami (talk) 08:14, 27 October 2015 (UTC)

The symbols are certainly not arbitrary. Why should they be arbitrary? We are not at liberty to invent new symbols. Inventing new symbols is pure WP:Original research. --mach 🙈🙉🙊 08:16, 27 October 2015 (UTC)

Symbols are arbitrary by definition. And of course we're free to invent new ones. Calling that OR is like calling infobox format OR. We can't provide RS's for the layout of our info boxes, but that has nothing to do with OR. OR deals with the infomational content of an article, not its presentation. — kwami (talk) 08:19, 27 October 2015 (UTC)

I don't care for your self-proclaimed definitions. Pronunciation transcriptions are different from infoboxes, tables etc. Pronunciation transcriptions are a part of the informational content of the articles, whereas infoboxes, tables etc. are merely a part of the presentation. The informational content of an article must not be WP:Original research. That is one of the WP:Core content policies. I know it is your personal POV that we are free to invent whatever transcription symbols we like. However, Wikipedia is not about your personal POV. Wikipedia is about WP:NPOV. --mach 🙈🙉🙊 08:31, 27 October 2015 (UTC)

Since you're intelligent enough to know your arguments are false, I take it you're not interested in debating in good faith if you think that won't get you what you want. I don't care to waste my time with bullshitters. — kwami (talk) 18:30, 27 October 2015 (UTC)

/u/ for /uː/ or /ʊ/ is not original research when /u/ is a weak vowel akin to /i/ (as in situation). It's exactly how Longman Pronunciation Dictionary uses this symbol. Peter238 (talk) 09:35, 27 October 2015 (UTC)

Sure, but that is not how it is used on this help page, where it is used for a stressed vowel that may either be /uː/ or /ʊ/, e.g. in the word roof. Such a usage of /u/ is probably OR. --mach 🙈🙉🙊 10:00, 27 October 2015 (UTC)

For a general discussion whether OR pronunciation symbols are acceptable, please see Wikipedia:No original research/Noticeboard#Is it OK for pronunciation symbols to be Original Research? --mach 🙈🙉🙊 10:22, 27 October 2015 (UTC)

Yeah. We might want to get rid of using /i, u/ for anything other than weak vowels. Peter238 (talk) 10:54, 27 October 2015 (UTC)

Now that's a legitimate discussion. Can we all admit that we're intelligent enough to understand the issues, rather than pretending to be idiots if we think that noise rather than reason will get us what we want? — kwami (talk) 18:30, 27 October 2015 (UTC)

I didn't realize we were using /i/ for non-weak vowels. Regarding /u/, my question is the same as I posed here; how many articles would be affected? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 09:15, 28 October 2015 (UTC)

I don’t know. I don’t think it’s relevant (though it sure would be interesting to know). The relevant question is: Are there any sources? --mach 🙈🙉🙊 09:46, 28 October 2015 (UTC)

For me it's relevant. We already have a handful of articles with terms that can't be presented in one transcription, even in a diaphonemic system. Since having such a number of transcriptions is unavoidable, I can see taking out some of these diaphonemes from our system and doing two pronunciations for these as a viable choice if the number is sufficiently small. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:41, 28 October 2015 (UTC)

I really don't think it's more than 50-100, probably less. I have no idea how to check that though. Peter238 (talk) 18:13, 28 October 2015 (UTC)

Plus we have precedence in articles like Iraq, where we already have variant transcriptions in spite of our OR diaphonemes. --mach 🙈🙉🙊 00:51, 1 November 2015 (UTC)

Yeah, that's pretty much what already said in my comment above. Am I missing something here or are you?— Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 07:01, 1 November 2015 (UTC)

"in spite of our OR diaphonemes": Mach still seems clueless as to what's going on, unless he's just being careless with his wording. The BATH vowel, the first diaphoneme he objects to, is not "OR". You can find hundreds of sources for it. — kwami (talk) 22:14, 1 November 2015 (UTC)

The issue that Mach has is with the use of one symbol for a difference of incidence. Is there a source that transcribes the BATH vowel differently than both TRAP and PALM? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 22:29, 1 November 2015 (UTC)

I believe we went over several when we introduced these distinctions. — kwami (talk) 03:00, 3 November 2015 (UTC)

I'm not seeing any in the archives. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 05:19, 3 November 2015 (UTC)

Idle claims. Sources have to be cited, not vaguely hinted at. --mach 🙈🙉🙊 08:14, 3 November 2015 (UTC)

That's my thinking, too. We could cite such a source at Help:IPA conventions for English — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:53, 3 November 2015 (UTC)

We don't need sources at all, unless the transcription is making an unsourced theoretical claim.

I just came across Flemming & Johnson (2007) for ⟨ɨ⟩ for the e of roses, but all the barred vowels were taken from specific sources when we introduced them. I suppose one might object to the more recent additions, the BATH and CLOTH vowels, if we cannot find sources that assign them a single symbol in the IPA, but there is no unsourced content here, just a novel transcription. — kwami (talk) 00:55, 4 November 2015 (UTC)

Please have a look at WP:VER. It says: “All material in Wikipedia mainspace, including everything in articles, lists and captions, must be verifiable. All quotations, and any material whose verifiability has been challenged or is likely to be challenged, must include an inline citation that directly supports the material.” There are no exceptions of the type you have claimed (“unless the transcription is making an unsourced theoretical claim”, “novel transcription”).

I guess that by “Flemming & Johnson (2007)” you mean Rosa’s roses: reduced vowels in American English. That is hardly relevant to our dictionary-style phonemic transcription system because it is a phonetic study without phonemic analysis. It never specifies what phoneme the sound [ɨ] belongs to, or whether it is the reduced allophone of different phonemes. It hints at the possibility that [ɨ] and [ə] could be the non-final and final allophones of the same phoneme. This means that the roses [ɹoʊzɨz] – Rosa’s [ɹoʊzəz] minimal pair could be analyzed as /roʊz#ɨz/ vs. /roʊzɨ#z/ (p. 94).

I am even more concerned about sources for the signs ⟨ʉ⟩ and ⟨ɵ⟩, though. --mach 🙈🙉🙊 01:58, 4 November 2015 (UTC)

According to your fundamentalist interpretation, we cannot have maps unless they are directly copied from a map in a RS. We cannot have tables unless they are directly copied from tables in a RS. We cannot caption an image unless the caption is taken from a RS. Those standards are for Wikisource, not Wikipedia.

Oh, and categories: we can't put articles in categories unless we have RS's that they belong there.

I take it that you have no substantive objection to our conventions? — kwami (talk) 03:12, 4 November 2015 (UTC)

Don’t be ridiculous. Categorization and original images are of course allowed. Have a look at the WP:Core content policies. --mach 🙈🙉🙊 08:16, 4 November 2015 (UTC)

I think the point was that you can't categorize someone in a way that is not supported by sources (e.g. adding a homosexual category to Barak Obama or whatever). It's not a good analogy, though. Our transcription systems are novel, synthetic systems originated at Wikipedia, and are WP-generated metadata, not subject to WP:RS as to their orthography. One is loosely based on IPA, the other loosely based on how American [DIK-shun-air-eez]] do it, and neither system is subject to any third-party, external verification. RS is satisfied as long as the pronunciation that emerges in the reader's brain is broadly correct (i.e., does not result in "coffee" being interpreted as pronounced "kerfer" or "coh-fee-eh"). Either of our transcription systems is purely a style of presentation of data, and is not the information itself, like our quotation and citation formatting templates, which are presentational processing of externally sourced data. How the word is said in mainstream varieties of English is the externally verifiable data in a pronunciation, not the encoding mechanism we used to convey it (even if one of them is based on an external standard). I agree we should actually more closely follow the standard, for several reasons I detailed at that NOR noticeboard thread, but "this violates NOR!" is no one of them. It's a broken argument.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  14:11, 12 November 2015 (UTC)

I don’t quite understand why you shy away from calling things by their name. “[N]ovel, synthetic systems originated at Wikipedia” is WP:Original research in a nutshell. Fortunately, the symbols of our broad IPA transcription scheme are for the most part based on relevant sources. I only have a problem with those symbols that are not. If you prefer I should argue for this on the grounds of common sense, so be it. I still think calling it original research is more concise.

A broad IPA transcription like /prəˌnʌnsiˈeɪʃən/ can introduce new, synthetic material on two different levels: (a) The pronunciation it represents might not be backed by relevant sources (e.g. */prəˌnjuːnsiˈeɪʃən/), or (b) the symbols it is composed of might not be backed by relevant sources (e.g. /prɵˌnʌnsiˈeɪʃən/). But you and others are saying that only (a) can be called original research, but not (b). Why do you think that? I do not understand it, and nobody has given me an answer yet. --mach 🙈🙉🙊 15:04, 12 November 2015 (UTC)

Noticeboard discussion

at Wikipedia:No_original_research/Noticeboard#Is_it_OK_for_Wikipedia_to_choose_its_own_pronunciation_symbols? — kwami (talk) 18:27, 27 October 2015 (UTC)

This has spilled over to Wikipedia talk:Manual of Style/Pronunciation#Use of ɵ for broad IPA transcriptions. I've suggested at both locations that this discussion should be re-centralized back here. I also note that two editors are making essentially unilateral changes to that guideline, despite the fact that the issue is not resolved. (I happen to agree with the edits, so I won't reverse them, but as a matter of WP:PROCESS this probably needs to be reined in until such time as a real consensus forms on this issue.)  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  14:00, 12 November 2015 (UTC)

The Diaphonemic System

I am finding it increasingly hard to follow this argument so, since my name has been mentioned in the Noticeboard Discussion, I would like to see if I have understood.

• Symbols: some people seem to feel that WP is in danger of departing from the principles of the IPA, whose symbols are used. I don’t see this as a problem: the IPA has devised the symbols, and when they are used in phonetic transcription each has a precise and literal meaning in terms of articulation. However, the IPA recognizes that the symbols may be used for many purposes (ref IPA Handbook p.3) and when they are being employed for a purpose other than recording precise phonetic properties the IPA does not attempt to limit how they are used. On the other hand, inventing completely new symbols would be outside the IPA’s principles. The present discussion is about the representation of English, and I don’t see any sign of anyone wanting to invent new symbols here.
• Symbols with diacritics: some people seem worried about the creation of new symbols by combining familiar symbols with diacritics, for example by adding a length mark ː to the symbol a to make aː. As far as I know this is perfectly OK. I do have doubts specifically about the creation of “barred” symbols, because I have always thought that the symbols ɨ and ʉ were one-off creations. The “barred ɪ” and “barred ʊ” symbols were, I believe, invented for the Oxford Dictionary of Pronunciation; the editors there refer in their Introduction to “the IPA convention of barring to signify centralization of high vowels and retraction of front vowels”, but that is not an IPA convention that I have seen referred to elsewhere. I certainly can’t claim to have read everything written about phonetic symbols, though.

• Choosing symbols for phonemes: here the situation is quite different. Symbols for phonemes do not have to be exactly linked to specific phonetic qualities, and a variety of criteria affects the choice of symbol. A familiar example is the symbolization of the CAT vowel, which can be /æ/ or /a/, both of which have their advocates. We still use the symbol /ʌ/ for the CUT vowel in spite of the fact that the present-day pronunciation of that vowel in RP/BBC and in GA is a long way from the phonetic quality [ʌ]. Although there is a general consensus around the transcription of RP/BBC, publications from Oxford University Press have departed quite significantly from that consensus in the choice of vowel symbols, as have some recent textbooks. Therefore I can’t see any problem with WP choosing symbols for English phonemes that are different from those used by established works
• Using phoneme symbols to represent more than one phonetic quality: phonemes may be pronounced in a number of different ways, depending partly on phonological context and partly on accent or dialect. For example, the diphthong symbolized /əʊ/ has a very wide range of realizations in British English accents and the symbol stands for all of these. Consequently it is understood that in a phonemic transcription any symbol may have a number of possible realizations and it is only possible to convert the phonemic representation into a plausible phonetic realization if one is in possession of the relevant contextual information. It’s not just a matter of the symbols themselves, but also of the analysis underlying them. For example, it’s perfectly possible to analyse the English vowel system as containing just 6 vowels. There is never a single correct phonemic analysis of a language or accent, unless one is one of the rare linguists who adopt a “God’s Truth” approach to the problem. So there is a great deal of leeway in the choice of phonemic analysis.

• Deviations from phonemic orthodoxy: in practical applications of phonetics and phonemes it is sometimes found desirable to make a deliberate violation of the rules of phoneme theory. The best-known example is the “happY vowel”, where in certain (unstressed) contexts the symbol i is made to stand for both /ɪ/ and /iː/. A parallel case is u, standing for /ʊ/ and /uː/. This might be called a supraphonemic representation. It was devised originally for use in dictionaries. We also find cases where phonetic information is added to dictionary phonemic transcriptions with the practical goal of helping users with pronunciation. In the LPD and the CEPD the “flapping” of American /t/ is shown by adding the voicing diacritic (subscript v) to the /t/ in words such as ‘better’, ‘getting’. This information is, strictly speaking, redundant (being predictable by rule). We could call this a subphonemic representation.

• Using one phonemic symbol to represent more than one phoneme: WP has adopted a transcription system which takes the principle behind the “happY vowel” and extends it further, with vowels a: and ɒː representing an either/or choice of vowel depending on speaker’s accent. This is where the main controversy seems to centre. I have to say that in my own dictionary work I never felt we could make such a scheme work. CEPD, LPD and ODP all give British (RP) and American pronunciations separately except where it is felt that there is no significant difference. Some time ago in this discussion I asked if there was any source which had actually used a diaphonemic system successfully, expecting the answer ‘no’, but Kwami pointed to the system devised for the Cambridge Grammar of the English Language (Huddleston and Pullum), which I had not been aware of. The system can be seen here (relevant passage is from p.13 onwards).

• Finally, two issues have to be considered:

a. is the diaphonemic system a case of OR?

b. can the general readership be expected to understand it?

In answer to the first, I think the CDEL case is a very good precedent to cite in support of WP using a diaphonemic system. In answer to the second, I seriously doubt if the public will understand how the system works. Faced with a transcription such as //baːθ//, almost anyone would take aː to signify a front vowel quality if they know anything about IPA symbols. I would tentatively suggest that a possible solution to all the argument could be to adopt the CGEL diaphonemic transcription completely in place of the current controversial system. This would mean that WP could continue to have a diaphonemic system for representing English, if that’s what people want, but it would be a system with impeccable credentials that could not be accused of being OR. RoachPeter (talk) 10:50, 4 November 2015 (UTC)

Thanks for your response, Peter. I don't blame you for not wanting to follow the entire conversation. You've provided some expert input, which is what I was hoping to get from you.

As it so happens, I recently came across a 1961 article that uses ⟨ᵻ⟩ dating back to ("System Status of Obscured Vowels in English" by Lee Hultzén in Language 37.4). I'm not sure exactly what it is intended to represent, though it doesn't seem to mean the same thing as the OED vowel means it. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 17:29, 4 November 2015 (UTC)

Hi Peter,

/a:/ is used for the back PALM vowel in quite a few sources. Generally, ⟨a⟩ can mean any low vowel in IPA transcription.

I just came across ᵻ again today for a central [ɪ̈]. And yesterday I found barred ⟨a⟩ for a central low vowel, contrasting with ⟨ä⟩ for a centralized (but still front) low vowel. That's informal, of course, but then so are ᵻ ᵿ. Though ᵻ ᵾ are quite established in Americanist notation. Maybe that's where the IPA usage comes from. — kwami (talk) 03:15, 6 November 2015 (UTC)

@RoachPeter: You are saying that “I can’t see any problem with WP choosing symbols for English phonemes that are different from those used by established works”, based on the fact that the newer “publications from Oxford University Press have departed quite significantly” from the “general consensus around the transcription of RP/BBC”. I think you have overlooked a decisive difference between Oxford University Press and Wikipedia: Oxford University Press has all the liberty of introducing new transcription symbols. Wikipedia, however, has chosen a strict WP:No original research policy which forbids introducing anything new.

You have referred to the Cambridge Grammar of the English Language (CGEL) transcription system as a justification of our diaphonemic symbols. However, the CGEL system does not go anywhere near as far as we do. To the contrary, when discussing – for example – the variance between /æ/ versus /ɑː/, the CGEL says that “instead of introducing a third symbol we give separate BrE and AmE representations when necessary” ([3], p. 16). That is the exact opposite of what we have been doing so far. --mach 🙈🙉🙊 08:33, 6 November 2015 (UTC)

Thanks for helpful responses. I am not competent to judge on OR issues, but from a common sense point of view it surely can’t be literally true that WP is “forbidden to introduce anything new”. To take an extreme example, if WP designed a different typeface for its own use, I can’t imagine that being OR. I was just trying to suggest that in the debate about the diaphonemic transcription, it is not helpful if people feel that the alternative is a set of phonemic transcription conventions that are set in stone. I take the point about the CGEL’s treatment of open vowels, but in general its transcription seems an acceptably pragmatic solution that doesn’t go too far from current practices. RoachPeter (talk) 12:03, 6 November 2015 (UTC)

I agree that the CGEL system would be a good solution, and an improvement over our current practices.

Coming back to your typeface analogy: Not every new typeface would be acceptable on Wikipedia. If we wanted to introduce a typeface where the character ⟨A⟩ looks like ⟨Ɑ⟩ – or where the character ⟨ɑ⟩ (as in the phoneme /ɑː/) looks like ⟨a⟩ –, then we would be introducing something new, and people would be right to challenge the introduction of such a new feature. --mach 🙈🙉🙊 13:06, 6 November 2015 (UTC)

There are already fonts that do the former, actually. People's challenging of this hypothetical font would be because of readability issues, not because of OR. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 14:58, 6 November 2015 (UTC)

Readability issues are one of the reasons that have been brought up repeatedly against the use of original research pronunciation symbols. When encountering a transcription such as /suˈdaːn/, a reader who knows the IPA is likely to conclude that /aː/ refers to the PALM vowel.

You have said yourself that “newness may be negatively impacting the readability of our system” [4]. I whole-heartedly agree with that. The negative impact of newness on readability sums up the whole point I have been trying to make in all these discussions. Maybe we can find common ground on these terms? The only reason why I have referred to WP:NPOV, V and OR is because I wanted to to justify the point that newness negatively impacts readability. --mach 🙈🙉🙊 17:11, 6 November 2015 (UTC)

I get your point, but this most certainly is not the case you have been making so far. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 18:14, 6 November 2015 (UTC)

That is your POV. To me, saying that material has newness that negatively impacts readability is the same thing as saying that material violates the WP:Core content policies. So from my POV, I have been making the same case all along. But wheter you call it newness or original research, it is good that we finally agree. --mach 🙈🙉🙊 19:17, 6 November 2015 (UTC)

Newness is not the same thing as original research. In the comment you were referring to, I was talking about newness to the readers. So even if we did find a source that transcribes the BATH vowel with different symbols from both PALM and TRAP, we would be alleviating nagging concerns about original research while the issue of readability persists. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:57, 7 November 2015 (UTC)

Because the WP:Core content policies “complement each other, they should not be interpreted in isolation from one another.” If there were a source for ⟨aː⟩ (WP:V), it would be no longer WP:OR in the strict sense, but our use of ⟨aː⟩ would still not be justified. Instead, we also have to take WP:NPOV into account: If a source for ⟨aː⟩ existed, it would probably represent a remote minority point of view. So in the strict sense, you are right that avoiding newness is not the same as WP:OR. In a broader sense, avoiding newness corresponds to the WP:Core content policies taken together (as they are supposed to be taken). Of course, the example with ⟨aː⟩ is hypthetical because for all we know, ⟨aː⟩ being used as a diaphonemic sign is indeed WP:OR. --mach 🙈🙉🙊 18:59, 7 November 2015 (UTC)

Talking about how exactly our core policies prompt us to reconsider the use of ⟨aː⟩ is academic wikilawyering. If we take a step back and think about it in the lens of common sense, the issue of readability or comprehension remains an issue. I have been a defender of ⟨aː⟩ in the system but I am reconsidering my support in light of points made by other users.

Insofar as a diaphonemic system incorporates what diaphoneme calls "diaphonic identifications" (that is, where speakers perceive sounds from different varieties as equivalent), it is helpful and IMHO intuitive; there might be some hiccups in guiding readers to the right approach of this system, but it works best when readers approach the transcriptions in the same way they might an orthography-based pronunciation system.

I've come to be a bit on the fence about how well a single symbol could work to represent the BATH vowel within the logic of the diaphonemic system; perhaps someone can point to studies that clarify the matter, but I suspect that the difference in incidence between /æ/ and /ɑː/ is as marked as that between /eɪ/ and /ɑː/ (in words like tomato). This makes BATH different in that it isn't an additional example of a phonemic contrast that not everybody makes but rather a difference in incidence (that is, a given set of words features one phoneme in one variety and another phoneme in a different variety) that even lay speakers would notice and thereby not make a strong diaphonic identification. Despite this reluctance, the mouseover feature of {{IPAc-en}} works to temper this concern as there is an anchor word (bath) that the reader can align their pronunciation to. However, if we extend ⟨aː⟩ (or whatever symbolization we choose) to any instance where there is variation between /æ/ and /ɑː/, then we lose the utility of the anchor word and thereby lose the necessary clarity for the system to work. A good example where we lose this is Denali, where the difference in incidence is actually the opposite of BATH (that is, /æ/ appears in British speech and /ɑː/ in American speech). — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 08:33, 8 November 2015 (UTC)

The same applies to ⟨u⟩ symbolizing the diaphoneme that can be either /uː/ or /ʊ/. Am I correct that we now agree about the following changes?:

• ⟨aː⟩ – no longer to be used
• ⟨ɒː⟩ – no longer to be used
• ⟨ɨ⟩ – to be replaced by ⟨ɪ⟩ which is used in newer OUP publications or in the CGEL
• ⟨ʉ⟩ – to be replaced by ⟨ʊ⟩ which is used in newer OUP publications

If that is correct, then the only new symbol we still disagree on is ⟨ɵ⟩. --mach 🙈🙉🙊 11:02, 8 November 2015 (UTC)

I wouldn't cry if we did those things, though I'd like to get a little more input from other editors on my reasoning before we make these changes. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 17:39, 8 November 2015 (UTC)

Very good then, so we have reached a consensus about the signs ⟨aː ɒː ɨ ʉ⟩. I will wait a few more days and then, if nobody has objected, make the requisit changes. The goal is working towards a removal of the OR alert. --mach 🙈🙉🙊 20:06, 8 November 2015 (UTC)

We actually don't have consensus on removing ⟨aː⟩ and ⟨ɒː⟩. Given the prior consensus, we can't assume that silence is agreement in this case, particularly from some of the editors who have already expressed their support of their use. For those two, I would like to see something more explicit. We may have more consensus to limit ⟨aː⟩ to the BATH vowel, rather than remove it outright. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 20:27, 8 November 2015 (UTC)

While silence is not agreement, it is not disagreement either. There was no prior consensus, but only a prior status quo. It was kept out of inertia in spite of many editors expressing their disagreement over the years. These discussions did not end in consensus, but because the defenders of the status quo were more persistent anwerers. --mach 🙈🙉🙊 22:28, 8 November 2015 (UTC)

Silence being neither agreement nor disagreement furthers my point that we should wait for other editors to contribute. The rest of your comment is flat out wrong. Discussions on the matter can be found here, here, and here. In those three discussions, there was absolutely no disagreement. The incorporation of ⟨aː⟩ and ⟨ɒː⟩ was implemented last year and, until recently, only one user argued against it as part of his general opposition to the diaphonemic nature of the system, which he opposes. He couldn't convince others to agree with him, which is why we didn't change our system. This charge of false consensus is really an untoward and uncivil attack on the character of myself and other contributors; I suspect, because of the gross inaccuracy of it, that you are making this implicit charge based on unscrupulous comments rather than a personal reading of the relevant archives. Be more careful about this. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 23:10, 8 November 2015 (UTC)

• Support for restricting the usage of ⟨aː⟩ only to BATH words removing the symbols ⟨aː, ɒː⟩ from our guide. Peter238 (talk) 23:14, 8 November 2015 (UTC)

I strongly oppose this. The use of ⟨aː⟩ as a symbol for the BATH vowel seems to be a new Wikipedia usage that has no precedents from outside Wikipedia. It would be misleading to readers because the length sign could make them believe that ⟨aː⟩ were a symbol for the PALM vowel.

@Ƶ§œš¹: What is that about? If I have offended you, please accept my apologies. It has not been my intention. My intention is just to oppose my POV to your POV. And believe me, I carefully read the past discussions about the OR issues. Most discussions did not end in consensus. --mach 🙈🙉🙊 00:39, 9 November 2015 (UTC)

I linked to the three discussions that led to the implementation of ⟨aː⟩ and ⟨ɒː⟩. How can you look at those and see that as no consensus? Who are the "many editors" who opposed their implementation? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 01:22, 9 November 2015 (UTC)

Editors who over the years have voiced their concerns about our pronunciation symbols include Taivo, Bazj, Fortnum, Kudpung, W. P. Uzer, and myself. The three discussions you have linked to do not show much disagreement, but they do not show strong consensus either. The issue was just not discussed controversely. The WP:NPOV, V and OR issue, for instance, was never discussed. RoachPeter’s concern that by introducing new diaphonemic symbols Wikipedia would “get too far away from contemporary practice” – a hint at the future NPOV/V/OR discussions – was not even answered. The reasoning was: Nobody is opposed, so let’s do it then. Afterwards, several users – including W. P. Uzer and me – have voiced strong disagreement. --mach 🙈🙉🙊 10:19, 9 November 2015 (UTC)

So far, nobody has disagreed with my above suggestion (removing ⟨aː⟩ and ⟨ɒː⟩, replacing ⟨ɨ⟩ and ⟨ʉ⟩ by ⟨ɪ⟩ and ⟨ʊ⟩, see [5]). In the spirit of WP:SILENCE, I am going to apply these suggestions soon.

Regarding previous consensus: Please consider that consensus can change – especially when previously unconsidered arguments have been raised (for instance, original research was not discussed in the previous discussions about ⟨aː⟩). I believe the previous consensus was really a wrongful consensus because it violates WP:NOR. Status quo stonewalling is not helpful. --mach 🙈🙉🙊 17:02, 11 November 2015 (UTC)

I'm sorry, but you do not have consensus. I oppose what you are doing. If you really have started this process, please stop. In addition, please do not misrepresent the conversation. Peter238 just stated that he disagrees with removing ⟨aː⟩, and your response saying that you disagree with him makes your lie quite blatant. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:46, 11 November 2015 (UTC)

Why do you oppose? What else do you propose? Why? Peter did not oppose my proposal; he made a different proposal. I am not misrepresenting the discussion. I will not give in to Status quo stonewalling. --mach 🙈🙉🙊 20:14, 11 November 2015 (UTC)

It's true that I don't oppose removing /aː, ɒː/. If I were to choose between keeping /aː/ and restricting its usage to BATH words, I would obviously choose the latter. That's what I meant. And if you're going to remove these symbols - by all means, go ahead. Peter238 (talk) 20:19, 11 November 2015 (UTC)

We have enough consensus to change ⟨ɨ ʉ⟩; I would recommend starting by altering {{IPAc-en}} to incorporate ⟨ᵻ ᵿ⟩ (but not ⟨ɪ ʊ⟩). As I said, I want input from more editors in regards to /aː, ɒː/. I think it's clear so far that we have a consensus to not have ⟨aː⟩ in non-BATH words (the extension of it to non-BATH vowels was most certainly done without consensus); if you want to go through and remove those, that would be fine but it might make for more work in the long run if we also decide to remove ⟨aː⟩ altogether. Give it time. Sometimes these things move slowly. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 21:08, 11 November 2015 (UTC)

• Since we seem to be voting, though nobody announced it: Support mach's proposal, removing //aː ɒː// and replacing //ɨ ʉ// with barred //ɪ ʊ// (not sure how to type those last symbols). In a previous discussion I supported the first two symbols, but I retract my support. I still understand the reasoning behind using them, but they're ripe for being misinterpreted by people who don't understand the concept of phonological comparison of dialects, as indicating that there are actually more low vowel phonemes in a given dialect than there actually are, and they have simply been entirely invented by Wikipedians. As for the second two symbols, it's weird to use the same diaphonemic analysis as the OED but use different symbols. It'll leave people wondering if the symbols mean the same thing as the slightly different symbols in the OED or not. Better to use the same symbols for the same analysis.

Comment: All these "consensuses" linked by Ƶ§œš¹ above have been unofficial agreements among the small number of editors who participated in the discussion, with no formal vote. I would propose that one of the more involved editors make this into a formal vote, listing the various proposals and allowing editors to choose between them (or perhaps rank the choices), so that it's clear which proposal is supported by which editors, and why. Then there would be no need for bickering over who has consensus.

Also, I'm sure there are editors who have opinions, but haven't posted because they don't want to deal with debating with the more vocal editors. Those editors, like me, might be more willing to vote. — Eru·tuon 22:01, 11 November 2015 (UTC)

Very good, so we all agree about ⟨ɪ⟩ and ⟨ʊ⟩. With regard to ⟨aː⟩ and ⟨ɒː⟩ I do not think polling would be a good idea. Any conensus that ends up in us using ⟨aː⟩ and ⟨ɒː⟩ is a wrongful consensus because there is no verification for such a use – it violates the core content policies. There is no verification either for the use of ⟨aː⟩ representing the BATH lexical set. Ƶ§œš¹ has tried to argue that pronunciation symbols are somehow exempt from verification, but he has failed to find any compelling reasons. His argumentation always boils down to some assumption he cannot justify any further, e.g. “[o]ur IPA symbolization is merely the presentation of facts”.

We do not need more discussion. The matter has been discussed at nauseam. Arguing that more and more and more and more discussion is needed is a typical tactic of status quo stonewalling. This help page has a long history of status quo stonewalling. This has to stop. --mach 🙈🙉🙊 07:57, 12 November 2015 (UTC)

mach: I don't know what to think of the OR thing. There's been so many words flung between you and Ƶ§œš¹ over whether the symbols are OR and whether it matters, that my brain has frozen over. I am unable to come to an opinion of my own. And it seems that most people disagree with you, so the best course of action would be to just say you disagree about the OR thing and stop talking about it. Agree to disagree. Focus your efforts elsewhere. Perhaps there are other reasons besides OR for Wikipedia not to use the symbols you dislike.

I'll ignore the thing about status quo stonewalling because I don't see how I'm supporting that. I want to end discussion by having a vote. And with a vote, there can be simple statements of why each person supports each proposal. No additional arguing over how wrong other editors think their reasoning is. — Eru·tuon 21:20, 12 November 2015 (UTC)

The symbols I am objecting to are new and unsourced. To some editors, including W. P. Uzer or myself, this is OR in a nutshell. To other editors, including Ƶ§œš¹ or SMcCandlish, it is not.

I think their argumentation is fallacious because they do not distinguish between OR on the transcription level (where the transcription scheme can be viewed as “style”) and OR on the single-symbol level. They are not denying new and unsourced material on the single-symbol level should be avoided, but they do not want to call it OR because transcription schemes are “style” on the transcription level – if I understand their argumentation correctly. By contrast, we think that any new and unsourced material should be called OR, whatever level it belongs to.

I am willing to agree to disagree whether new and unsourced material on the single-symbol level is called OR or not – as long as the new and unsourced material is removed. --mach 🙈🙉🙊 09:32, 13 November 2015 (UTC)

I changed my vote. While I don't really care whether we restrict ⟨aː⟩ to BATH words or remove it, we really have to remove ⟨ɒː⟩. The reason is that the CLOTH vowel in General American has a different (wider) distribution than in conservative RP, so that while the word cloth is pronounced /klɔːθ/ in both conservative RP and non-cot-caught-merged GA (and /klɒθ/ in contemporary RP), the word dog is pronounced /dɔːɡ/ in NCCM GA, but /dɒɡ/ in all varieties of RP! This is too confusing. Peter238 (talk) 18:29, 12 November 2015 (UTC)

• Comment On a purely practical level, we cannot use Mach's proposal of soft-formatted ⟨ɪ⟩ and ⟨ʊ⟩ to replace ⟨ɨ⟩ and ⟨ʉ⟩ (though I just came across a source last week that used ⟨ɨ⟩), we need to use proper Unicode ⟨ᵻ⟩ and ⟨ᵿ⟩ so that they're copy-friendly.

Also, Mach's contention that any consensus he disagrees with is the wrong consensus is absurd. Also interesting that he's complained about a small cabal having decided things, but now that he's part of the discussion he wants the decision to be made by a small cabal. We may get rid of the CLOTH and BATH vowel, but it needs to be through consensus, not diktat. — kwami (talk) 20:15, 12 November 2015 (UTC)

I think ⟨ᵻ, ᵿ⟩ work just as well as ⟨ɪ, ʊ⟩. I do not know where you take the copy-friendliness requirement from – what is your basis for this argumentation? But anyway, if you prefer ⟨ᵻ, ᵿ⟩, that is fine with me.

As I said [6], the source you found for the use of ⟨ɨ⟩ is not relevant because it is not a broad phonemic transcription, but a purely phonetic study without any phonemic analysis (unless I guessed wrong and the source you only hinted at is not the one I identified – then please identify your source).

Refuting kwami’s straw men: I have not “complained about a small cabal having decided things”. I have complained about a small group of users engaging in status quo stonewalling. I have not said that “that any consensus he disagrees with is the wrong consensus”. What I have said is that a consensus that violates the core content policies is a wrongful consensus. Nobody wants to get rid of the CLOTH and BATH vowels – we want to get rid of the signs ⟨aː⟩ and ⟨ɒː⟩ because they are new and unsourced (which some would call OR, but some wouldn’t).

It seems we are reaching consensus:

• ⟨aː⟩ – no longer to be used
• ⟨ɒː⟩ – no longer to be used
• ⟨ɨ⟩ – to be replaced by ⟨ᵻ⟩ which is used in newer OUP publications or in the CGEL
• ⟨ʉ⟩ – to be replaced by ⟨ᵿ⟩ which is used in newer OUP publications

--mach 🙈🙉🙊 09:06, 13 November 2015 (UTC)

Except "I strongly oppose this", and "it seems that most people disagree with you", and "We may have more consensus to limit ⟨aː⟩ to the BATH vowel, rather than remove it outright", and "contention that any consensus [one] disagrees with is the wrong consensus is absurd", and "On a purely practical level, we cannot use Mach's proposal", and etc., etc. I detect a strong current of WP:ICANTHEARYOU going on. The nom's comment "All these "consensuses" linked ... above have been unofficial agreements among the small number of editors who participated in the discussion, with no formal vote" does not properly reflect WP:POLICY and WP:CONSENSUS. Every consensus on WP is determined by whoever shows up, there is no such thing as "official" on WP, and consensus does not require an RfC (though one often helps arrive at consensus), and RfCs and other discussions are not a "formal vote". There are not formal votes on WP, other than for elections of a few sorts.

A long-standing consensus, however few editors formed it initially, gains in consensus level the longer it stands and the more it is adopted in actual practice across a range of WP topics. Thus a sudden putsch to change it, especially if it may affect a large number of articles, is expected to gain a significant level of editorial buy-in before consensus can be said to have actually changed. This (admittedly flawed) pronunciation transcription system has been in place and stable for a long time, and the sky did not fall. There is no hurry, and we should get it right this time. I would strongly suggest an actual RfC. I think it should be either be written by someone other than Wust, or by him if he agrees to avoid making NOR arguments in it, which he knows that several of us have challenged (and provided details rationales for challenging, to which he's basically responded with a combination of "I don't get it" and "I just don't agree" without a counter rationale. At the WP:NORNB discussion, Wust conceded that this would not be settled at a NOR matter, but a WP:COMMONSENSE matter, so let's stop clouding the issue. Let's also stop pretending this discussion, which is part of a three-way discussion split, the exact opposite of discussion centralization, can possible represent a consensus change to major way-things-are-done (even if it's not actually a guideline or policy itself), and let's instead open a proper, organized, centralized discussion about it via RfC, which will draw in additional editors via WP:FRS. (NB: I say all that as someone who also opposes what has been done with quasi- and pseudo-IPA here; but it does the goal of rectifying that problem no good at all to rush through a proposal, that is roundly or in-detail opposed by so many, to try to impose a false consensus that will just get reverted. It would do more harm than good, since it would give a "not this crap again" excuse to ignore a later, more WP:PROCESS-cognizant attempt to resolve the issue.)

PS: An RfC on this should probably enumerate each of the problematic symbols, what the problem(s) is/are with each, and what the known alternative approaches are, as well as an overview of the problem for people not very familiar with IPA.  — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼  11:36, 19 November 2015 (UTC)

Your quotes are out of context. A closer look shows that not a single one of them serves as evidence for lack of conensus:

• “I strongly oppose this” [7] – that was not against the consensus we are forming here, but against an alternative proposal. And, by the way, it was me who uttered it.
• “it seems that most people disagree with you” [8] – that was not against the consensus we are forming here, but that was Erutuon’s remark about the discussion at WP:NORNB.
• “We may have more consensus to limit ⟨aː⟩ to the BATH vowel, rather than remove it outright” [9] – this was indeed about the consensus we are forming here. We continued to discuss. I gave my reasons why I think limiting ⟨aː⟩ to the BATH vowel is not a good idea [10]. Two editors explicitly agreed that they favor dropping ⟨aː⟩ entirely [11] [12]. No editor stated any opposition – not even kwami or Ƶ§œš¹. It seems we have found a consensus about this point.
• “contention that any consensus [one] disagrees with is the wrong consensus is absurd” [13] – that was not against the consensus we are forming here, but a straw man argument by kwami. I had never said such a thing.
• “On a purely practical level, we cannot use Mach's proposal” [14] – this was indeed about the consensus we are forming here. Even while I do not follow kwami’s reasoning in this regard, I have altered my position [15]. It seems we have found a consensus about this point.

So, all things considered, it appears we have reached a consensus in this discussion.

Your interpretation of WP:CONLEVEL seems one-sided to me. You are setting a very high bar for the removal of pronunciation signs, but these same signs never passed such a high bar when they were introduced. --mach 🙈🙉🙊 13:43, 19 November 2015 (UTC)

BATH vowel symbol: pros and cons

@DavidPKendal: You have written: “I think it [the symbol /aː/] should be restricted only for use in BATH words” [16]. I am taking the liberty of answering here, because this is where we have discussed removing /aː/ (and removing /ɒː/ and replacing ⟨ɨ ʉ⟩ by ⟨ᵻ ᵿ⟩). Also, it was you who (re)proposed the symbol (see Help talk:IPA for English/Archive 12#Proposal for new symbols) and added it to the help page [17] (sorry, I had not looked that up earlier or I would have pinged you).

I think inventing our own symbol for the BATH vowel is not a good idea. My main reason is that there is no precedence. Basically, I think it is original research (some editors agree to this, but others don’t). It is true that our broad phonemic transcription scheme is somewhat different from others because we intend to be pan-dialectal (or at least multi-dialectal). But we are not the only ones with this intention. Another example is the CGEL. I actually think their solution is even better than ours because they are not as decidedly rhotic as we are – which means their scheme has more NPOV.

Now with regard to the BATH vowel, the CGEL argues that both British and American English have only two low vowels: the TRAP vowel and the CALM vowel (if I am not mistaken, these two vowels are distinguished in most dialects of English, though some might have additional low vowels). The BATH vowel is not a third low vowel different from the other two. Instead, it is identical to one of them. To which one – that depends on the dialect. Therefore, the CGEL says “instead of introducing a third vowel we give seperate BrE and AmE representations when necessary” (CGEL. “Preliminaries”, p. 16). I think we should follow this example instead of introducing a new symbol.

Also, it seems quite counterintuitive to me that some words where the pronunciation varies between /æ/ and /ɑː/ should be written with a special sign (the BATH lexical set), while others should not (words like Sudan or Iraq). The distinction is actually quite intricate and requires knowledge about the diachrony of English vowels. Furthermore, I guess that among the words on the English Wikipedia where the pronunciation varies between /æ/ and /ɑː/, most do not belong to the BATH lexical set. The pronunciation of the BATH lexical set words is usually trivial, so they do not get an IPA transcription. Foreign or complicated words such as Sudan, Iraq or Nazism are more likely to get an IPA transcription, but these words do not belong to the BATH lexical set proper. So there would be very few proper BATH words left – too few for justifying the invention of a new symbol. --mach 🙈🙉🙊 17:59, 21 November 2015 (UTC)

"requires knowledge about the diachrony of English vowels" — false, and doubly false in the context of Wikipedia. Since our pronunciations have to be referenceable, use of /aː/ should be clear from our sources. Unfortunately I don't have a copy of the LPD to hand at the moment, but looking at the front matter in Amazon Look Inside, I'm reasonably certain that we could use it for BATH vowel symbols, using it iff both /æ/ and /ɑː/ are listed as variants in the British section, and the /æ/ variant is marked with the § 'BrE non-RP' symbol. (See p. xix.) Other dictionaries like the EPD or the Oxford pronunciation dictionary may also mark this; I haven't seen them. As a more broadly referenceable rule, any occurrence of /æ/ in AmE/northern BrE in these contexts, where standard pronunciation and general dictionaries (esp. of British English) give both /æ/ and /ɑː/.

Some dialects do have a three-way TRAP–BATH–PALM. See Accents of English vol 2 p 346–7.

Finally, I wish you'd get off this "original research" hobby horse of yours. It is well settled that, as a help page establishing a notational convention, the IPA key is not subject to Wikipedia's normal proscriptions on original research, any more than the Manual of Style is. Nor is WP:USdict or WP:PRK (it's worth noting that this page was originally in the 'Wikipedia' namespace like those two, and I personally have quibbles with the decision to move it into the Help section). DavidPKendal (talk) 18:37, 21 November 2015 (UTC)

OK, so diachronic knowledge is not necessary, but knowledge on how to handle specific indications about British and American pronunciations. But couldn’t there be false positives (an almost false positive is oxfordlearnersdictionaries.com – Iraq)? The knowledge about specific phonetic contexts is pretty much what I meant when I referred to knowledge about diachrony.

So there are dialects that have a distinct BATH vowel. That is interesting – a pity there is nothing about this yet in Wikipedia. Are these dialects noteworthy enough so we should use them in our transcriptions (keeping in mind that very few words are concerned and that we do not encode all distinctions, see Help:IPA for English#Dialect variation)? Is there a broad phonemic transcription for these dialects?

If it is “well settled that, as a help page establishing a notational convention, the IPA key is not subject to Wikipedia's normal proscriptions on original research” – then where has this ever been settled?

Calling the invention of new pronunciation symbols original research is certainly not my “hobby horse”. Other users have shared this POV – see e.g. [18], [19], [20], [21], [22], [23], [24], [25], [26]. Your are trying to make it look as if I were the only one to uphold this POV, even though I am demonstrably not the only one. It would serve your POV better if you would try to justify it. Trying to make our POV look bad does not serve your POV. The justification for our POV is very simple: Creating new material is original research; coining new uses for IPA characters is creating new material; consequently, coining new uses for IPA characters is original research. But I do not even need to justify this POV because Wikipedia policy says that the burden of proof is not on my side when I raise the original research issue.

The comparison to the respelling keys is not a good justification (and why do we have two different ones?). They are unsystematic by nature, and what is more imporant, they have been equally accused of being original research (see e.g. [27], [28], [29], [30]). I have not yet seen any substantial justification why coining new pronunciation symbols should be exempt from WP:NOR. The best justifications I have seen stop at the observation that a pronunciation symbol scheme can be viewed as a pronunciation’s style (opposite the actual pronunciation as the pronunciation’s content). They then jump to the conclusion that the coining new pronunciation symbols cannot be original research. --mach 🙈🙉🙊 21:33, 21 November 2015 (UTC)

Reduced and Variable Vowels

I hate to say this, but the problems with “variable” and “reduced” vowels have still not been fixed. If we consider the “happY” vowel, we find it in the “Reduced vowels” box, with ‘happy’ and ‘serious’ as examples. Published works which make use of the "/i/-for-happY" convention also use /u/ for a corresponding back rounded vowel. However, the /u/ symbol appears here in the “Variable vowels” box. The final example, ‘situation’ is typical of the words used to exemplify the use of /u/, but the other two words, ‘bedroom’ and ‘roof’ are quite different – these are examples of a Wikipedia diaphoneme, realized either as /ʊ/ or as /u:/. All the other vowels in the “Variable vowels” box are restricted to unstressed syllables, as are all the vowels in the “Reduced vowels” box, so /u/ (which can be stressed in ‘roof’, ‘room’) stands out as different. If asked to suggest a solution, I would say remove the heading “Variable vowels”, merge the two boxes and class all of these as “Reduced vowels”. Then /u/ should be used only for unstressed back rounded vowels. RoachPeter (talk) 15:05, 21 November 2015 (UTC)

I can get behind that. My arguments against using /u/ are similar to those against extending /aː/ to non-BATH vowels that I said above: {{IPAc-en}} utilizes signal words to help readers understand which word in their own speech uses this vowel but it doesn't seem like bedroom or roof are part of a class of words that vary in a consistent way like BATH does. Without that consistency, there's no accurate way for readers to get which pronunciation they should be producing. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 15:18, 21 November 2015 (UTC)

I agree we should restrict ⟨u⟩ to weak syllables. The use of ⟨u⟩ for ‘diaphoneme /uː/ OR /ʊ/ (in stressed syllables or in weak syllables)’ is a new and unsourced symbol invented on the English Wikipedia – what I (and others) call original research. In previous broad phonemic transcription schemes, it is only used for ‘weak syllable vowel that can be analyzed as /uː/ or /ʊ/ (exclusively in weak syllables)’.

It was DavidPKendal who has suggested this symbol on this talk page – citing Wells and providing the example words situation and bedroom (see Help talk:IPA for English/Archive 12#Proposal for new symbols). Since nobody had objected, DavidPKendal included ⟨u⟩ later on the help page (with the example words situation and bedroom [31]. Then, Peter238 suggested the additional example word roof on the help page, where the sound is unmistakingly in a stressed syllable (see Help talk:IPA for English/Archive 12#Variable vowels, and he added it on the help page [32]. The point that Wells only uses the sign exclusively in weak syllables was never raised.

Also, the use of ⟨ɵ⟩ in a broad phonemic transcription scheme seems to be our own invention. Unfortunately, the reference to Bolinger (1989) is not complete. It lacks a page number. I have searched for the mention in the limited preview on Google Books, but I have not found it. However, from what I have seen it seems unlikely that Bolinger would propose a broad phonemic transcription scheme – his book on intonation and speech melody has a very different purpose. But even if he had proposed a broad phonemic transcription scheme, I think we should stick to more mainstream schemes (think WP:NPOV), such as the ones mentioned on the incomplete list of Help:IPA conventions for English. --mach 🙈🙉🙊 16:11, 21 November 2015 (UTC)

Indeed, /u/ should only be used in weak syllables. Also, in adding /aː/ I intended it to be used only for the BATH vowel, but it seems to have found its way into other words which have any kind of /æ~ɑː/ variation. I think it should be restricted only for use in BATH words. DavidPKendal (talk) 16:16, 21 November 2015 (UTC)

Yep, I probably shouldn't have suggested using /u/ in stressed syllables. It wasn't the best idea I think. Peter238 (talk) 16:22, 21 November 2015 (UTC)

Thanks for summarizing the talk page discussions, Mach. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 17:52, 21 November 2015 (UTC)