Help talk:IPA/English/Archive 20

Pronunciation of equation
This is far from a big deal, and I may have been the only one to notice it, but just thought I would check we are in consensus on removing equation as an example of /ʒ/, reason being some pronounce it /-ʃən/. Rovingrobert (talk) 11:04, 27 May 2017 (UTC)


 * Yes, I agree.   D b f i r s   14:47, 27 May 2017 (UTC)
 * Some people pronounce erosion /-ʃən/ too. To what extent can we account for alternative pronunciations? Rovingrobert (talk) 05:42, 28 May 2017 (UTC)
 * Erosion is non-standard with /-ʃən/ (though I have heard it) whereas the /-ʃ/ in equation is a genuine alternative according to the OED. For examples of the /ʒ/ sound, it would be ideal to choose a word that has only one pronunciation (at least for 99.9% of speakers of the language, there will always be someone who mispronounces any word we choose).    D b f i r s   06:10, 28 May 2017 (UTC)
 * Couldn't agree more. Rovingrobert (talk) 07:19, 28 May 2017 (UTC)
 * There's no need to replace it. As for erosion we already have vision for the same kind of environment (spelled "-sion" and pronounced ), and three words is already enough to illustrate what the sound is. Nardog (talk) 08:48, 28 May 2017 (UTC)
 * Cool. I was also interested in discussing how /n/ is sometimes pronounced as /m/ in words like rainbow and sandwich. Rovingrobert (talk) 11:34, 28 May 2017 (UTC)
 * Coda nasals typically assimilate to the following sound across languages, i.e. English and think with [ŋ], inpatient with [m], etc. Now whether these sounds belong to the same phonemes as onset nasals or to a different archiphoneme is up to debate (or a matter of preference). So dictionaries disagree whether to transcribe them as /n/ or /ŋ, m/ (it seems [ŋ] is typically /ŋ/ and [m] is typically /n/, I don't know why). Nardog (talk) 12:28, 28 May 2017 (UTC)
 * Maybe it's worth a cite note. By the way, the same phenomenon of /n/ going to /m/ seems to happen in unbelievable. Rovingrobert (talk) 09:48, 29 June 2017 (UTC)

Pronunciation of /ɒ̃/ as /m/
The cite note for /ɒ̃/ states that it is only found in French loanwords and often replaced by /n/ or /m/. I've never heard it replaced by /m/ myself; is this a known alternative? Rovingrobert (talk) 10:09, 10 June 2017 (UTC)
 * Perhaps would be added if the nasal vowel came before a labial consonant, like in our word somber or sombre, which originally came from French. I mean, there the French nasal vowel is even represented with m. — Eru·tuon 21:37, 12 June 2017 (UTC)
 * Makes perfect sense; thank you! Rovingrobert (talk) 10:20, 13 June 2017 (UTC)
 * I'm not sure if sombre is a good example to include in the key because /ɒ̃/ mostly occurs in more recent loans (dictionaries only list /m/ for sombre, whereas they do list both some kind of nasal vowel and /n/ as variations for bon vivant). It could also be [ŋ] or even [ɱ] depending on the environment, so the note might be better as just something like "replaced by a vowel and a nasal consonant" ("pronounced as" rather than "replaced by" might be even more accurate, since the phonemic status of /ɒ̃/ is not clear in the first place). Nardog (talk) 17:18, 13 June 2017 (UTC)
 * Right, somber was not an example of a word that would ever be pronounced with a nasal vowel in English, and I'm not advocating that it be added as an example on the main page. — Eru·tuon 22:41, 18 June 2017 (UTC)
 * Don't count anything out in such an international language. Aside from a number of malapropisms, there are also concepts like Franglais. Rovingrobert (talk) 09:54, 29 June 2017 (UTC)

Should we stop distinguishing /i, u/ from /iː, uː/ and/or stop transcribing "length"?
Analyses with a separate /i/ include, for instance, the ones by the current EPD or by Longman. --mach &#x1f648;&#x1f649;&#x1f64a; 05:15, 19 June 2017 (UTC)
 * Except that they don't. Both of them use the symbol $\langle\rangle$ to denote a free variation (sort of) between and . It's one of several non-phonemic symbols that they use. What I meant was analyses that treat  as a vowel separate from those two. And as you know, pronouncing HAPPY with  is quite obviously old-fashioned in both RP and GA. Mr KEBAB (talk) 11:34, 19 June 2017 (UTC)
 * I think I'll second Mach's opposition to deprecating the length marks simply on the grounds that that's what I suspect most people are used to seeing. Like it or not, most of the IPA-based transcriptions of English words encountered daily (namely on dictionaries) are more or less based on the traditional RP-based system (even when they're not of RP). (Apparently $⟨i⟩$ was "widely used [to represent /ɪ/] in EFL work in the first three-quarters of the twentieth century" (:9), too.) In the end, IPA on Wikipedia must be or at the very least resemble what most readers are familiar with, in honor of WP:LEAST (same reason I'm opposed to /ᵻ, ᵿ/).
 * As for discontinuation of /i, u/, while I agree the lax variants of word-final /i/ and prevocalic /u/ are probably negligible at this point and better merged with /iː, uː /), I'm ambivalent about prevocalic /i/. Media being transcribed as /ˈmi:di:ə/ or /ˈmi:di:.ə/ as opposed to /ˈmi:diə/ somehow looks very off; but that might be just me. Nardog (talk) 22:43, 19 June 2017 (UTC)
 * For length, see . It wouldn't cause a lot of confusion and it wouldn't be original research. There already are (American) dictionaries that use it, and I believe Mach has a list of them (it's somewhere on WP).
 * As far as the media thing is concerned, what you're thinking about is compression. Your usual pronunciation of that word is probably, with an approximant rather than a vowel, and that is the source of your confusion. If it's not, then you must be taking the length mark too literally, as the actual length of English vowels varies considerably from environment to environment. It's perfectly understandable that the first in  would be longer and the second one shorter. Not only is it unstressed, it also appears immediately after the stressed syllable. This is one of the reasons I wouldn't use length marks. They can be just as confusing. No system is perfect. Mr KEBAB (talk) 00:06, 20 June 2017 (UTC)
 * I just checked Longman Pronunciation Dictionary (2008) and Cambridge English Pronouncing Dictionary (2011). The RP recordings say, the GA recordings say . The difference I'm hearing between the first and the second is definitely caused by stress. Mr KEBAB (talk) 00:10, 20 June 2017 (UTC)
 * Well, /jə/ is the same as /ɪə/. Note that RP has /ɪə/ as a phoneme while GA doesn't, so in RP [ɪə] can be transcribed phonemically as /ɪə/ even when /ɪ/ is considered checked, whereas in GA one has to write /jə/ to transcribe [ɪə] only with phonemes (compare ). So in media, we have a classic example of a word where a diaphoneme splits into /ɪ/ in RP and /i/ in GA, which is all the more reason to preserve prevocalic /i/ as a diaphoneme (unless we ditch the length mark, obviously). Perhaps we should acknowledge /iə/ as a distinct diphthongal diaphoneme before deprecating /i, u/ altogether. (If so, /uə/ (and perhaps /juə/) should also be acknowledged.)
 * Going back to the length marks, I still don't see enough reason because, in addition to what I said above, /iː, uː/ are diphthongal in many accents anyway, and more importantly, few languages contrast [i, u] and [ɪ, ʊ], so they are more intuitive to those familiar with IPA notations of other languages too. Wells even says it's "a de facto standard" and "equally 'scientific'" in the article you cited.
 * Regardless, I'd say for now we should propose the discontinuation of /ᵻ, ᵿ/ first, then move on to the rest of our suggestions (making changes is easier when done gradually). Especially dropping the length marks visually affects many articles so I'd like to hear opinions from others. Nardog (talk) 02:03, 20 June 2017 (UTC)
 * In that word, where OD writes LPD and CEPD would write  (with the non-phoneme  denoting the  variation, with the latter variant being old-fashioned). The phoneme  is another story, and in contemporary RP  has little to do with the sequence.
 * I'm not sure what the diphthongal quality of has to do with this. Both  and  denote monophthongs in strict IPA, and we're talking about phonemic transcription. Also, there are dialects in which  are always monophthongal or only very slightly diphthongal. This includes Scottish English, South African English and at least some Welsh English.
 * I'm fine with putting this conversation on hold. Mr KEBAB (talk) 02:19, 20 June 2017 (UTC)
 * Are you sure /ɪ/ is outdated even word-internally? You said LPD and CEPD say [jə] and write /iə/, and that is precisely the same as [ɪə] since /j/ is the semivocalic counterpart of /ɪ/. Maybe you're right in that /ɪə/ has little to do with /jə/ since OED still writes the HAPPY vowel with /ɪ/. So I take back what I said about recognizing /iə, uə/ as diaphonemes and instead say that we should preserve /i/ at least word-internally, because if we wrote /i:ə/, that would be inaccurate in accents that say [ˈhæpiː] yet [ˈmiːdɪə]. (Which makes yet another reason I think we shouldn't ditch the length marks because, if we did, /iə/ would imply the same thing as the inaccurate /iːə/.) We would also need evidence as to whether prevocalic /ʊ/ is outdated or not should we drop /u/ as well.
 * Also, just curious, how would you say we should transcribe media, mediocre, superfluous, etc. if we dropped the archiphonemic /i, u/ but not the length marks?
 * I brought up the diphthongality (if that's a word) simply to make the case that /iː, uː/ are often perceptibly longer than /ɪ, ʊ/. Since the diaphonemic notation is used widely across the site, whose most readers are linguistically uneducated and/or ESL/EFL speakers, it is reasonable if not important to reckon with intuitiveness as well as the mere phonemicity of the segments. Nardog (talk) 02:28, 22 June 2017 (UTC)
 * These blog posts by Wells make a case that there are prevocalic high front vowels that could be either semivocalic or nuclear, yet are phonologically distinct from /j/. Let's keep /i/ (that is distinct from /iː/), at least word-medially. Nardog (talk) 13:13, 22 June 2017 (UTC)
 * Okay, Wells is an authority, but he's more of a phonetician than a phonologist, and I don't see very clear reasoning in favor of a separate phoneme. The posts just say that, for some reason, there is variation between and  in some words and not others. That is not, by itself, evidence for a separate phoneme. The use of a separate symbol for varisyllabic  seems to be for convenience – easier to use a single symbol than to write two separate transcriptions,  and  . If there is phonological reasoning that I am missing, please point it out. — Eru·tuon 19:11, 22 June 2017 (UTC)
 * If I'm not mistaken, it sounds like you might have actually missed the part where I said "I take back what I said about recognizing /iə, uə/ as diaphonemes and instead say that we should preserve /i/ at least word-internally". I've struck the part I retracted so it would be obvious. Nardog (talk) 21:31, 22 June 2017 (UTC)
 * But I don't understand: if are not diaphonemes, then where would you use  word-internally? — Eru·tuon 22:28, 22 June 2017 (UTC)
 * /i/, obviously. So to recap, Mr KEBAB has suggested that we use /i/ for both FLEECE and HAPPY, where I suggest we use /iː/ for both FLEECE and HAPPY, while retaining /i/ for the word-medial prevocalic high front vowel that is transcribed variably as /ɪ/, /i/, /j/. Nardog (talk) 13:06, 23 June 2017 (UTC)
 * Sorry I might have misread your question. I first suggested that, if we deprecate /i/ (for HAPPY), we should add /iə/ as a diaphoneme. But now I argue, keep /i/ (because there are words like mediocre where /i/ would have to appear not before /ə/). Nardog (talk) 13:22, 23 June 2017 (UTC)
 * Ahh, I see what you mean now. I guess I thought you were proposing getting rid of and using at the same time, which was weird. — Eru·tuon 17:20, 23 June 2017 (UTC)

...I'm not sure if I follow you two anymore. But anyway:

Are you sure /ɪ/ is outdated even word-internally? You said LPD and CEPD say [jə] and write /iə/, and that is precisely the same as [ɪə] since /j/ is the semivocalic counterpart of /ɪ/. - The transcription indicates that it can be pronounced. is the same as, is close to it but not the same. is the approximant counterpart of, a fully (or almost fully) close front unrounded vowel. English is typically considerably lower and less front,  or something a bit higher.

(...) and instead say that we should preserve /i/ at least word-internally, because if we wrote /i:ə/, that would be inaccurate in accents that say [ˈhæpiː] yet [ˈmiːdɪə]. - And what are those accents? To my ears, said by a younger (understand this term liberally) RP/GA speaker would sound obviously old-fashioned (I was going to say non-native, but that'd be an exaggeration). You might be confusing something here, but I'm not sure.

We would also need evidence as to whether prevocalic /ʊ/ is outdated or not should we drop /u/ as well. - Agreed. That post from Geoff Lindsey's blog that I linked to is a good start.

(...) how would you say we should transcribe media, mediocre, superfluous, etc. if we dropped the archiphonemic /i, u/ but not the length marks? - Like this:

I'm not sure whether Wells considers to be phonemes. I really don't think he does.

Keeping word-medially is not a good idea. We should follow what dictionaries do or stop distinguishing from  altogether. Unless we want people to be confused? And, again, we're already inconsistent in this. We transcribe California as. The editor failed to realize that represents the FLEECE vowel that can be compressed into an approximant. Mr KEBAB (talk) 13:35, 23 June 2017 (UTC)

The LPD may not be the best source in certain (rare) cases. See. The CEPD transcribes believe with. Mr KEBAB (talk) 13:57, 23 June 2017 (UTC)
 * I agree that Wells does not consider to be phonemes. They're just symbols for notational convenience. And apparently  means different things in words like city than it does in words like media. City has, while media has . Kind of confusing. — Eru·tuon 17:20, 23 June 2017 (UTC)
 * While Wells does not openly advocate for an /i/ phoneme, it appears he sympathizes with a polysystemic analysis, where /i/ is a phoneme. In the polysystemic approach, there is a strong vowel system where /iː/ contrasts with /ɪ/, and a weak vowel system where /i/ contrasts with /ə/. Wells says “there is a good case” for such an analysis and he regrets that “[u]nfortunately” it is difficult to grasp. --mach &#x1f648;&#x1f649;&#x1f64a; 19:01, 23 June 2017 (UTC)
 * I suppose that's similar to analyzing Russian vowels as (strong) and  (weak). In each system, the vowels contrast with each other in that system but not with the vowels from the other system. But I'm still not sure whether we need that level of detail here. We don't really aim at an in-depth phonological analysis of English words. Mr KEBAB (talk) 19:48, 23 June 2017 (UTC)
 * I thought /j/ was the counterpart of /ɪ/ in the phonology of English, even though phonetically (or in languages where there's no /i/–/ɪ/ contrast) it is [i]–[j]. LPD1 says:
 * Rather than changing all the way to the corresponding semivowel /j/ (giving /ˈliːn.jənt/), the /i/ may merely come to form the less prominent part of a crescendo diphthong /ĭə/, thus /ˈliːn.ĭənt/. Similarly, influence /ˈɪn.flu‿əns/, rather than becoming /ˈɪn.flwəns/, may be pronounced with a crescendo diphthong /ŭə/, thus /ˈɪn.flŭəns/. [p. 175]
 * If /j, w/ were the counterparts of /i, u/, the above quote makes little sense.
 * If you consider /j/ the counterpart of /i/, then, when you wrote LPD and CEPD say /ˈmidjə/, you meant something like [ˈmiːdiə] or [ˈmidĭə]?
 * So read the second quote you gave as "We should preserve /i/ at least word-internally, because if we wrote /i:ə/, that would be inaccurate in accents that say [ˈhæpiː] yet [ˈmiːdiə] (alternatively: [ˈhæpi] yet [ˈmidĭə])" instead.
 * Keeping /i, u/ word-medially is not a good idea. – Why?
 * We should follow what dictionaries do or stop distinguishing /i, u/ from /iː, uː/ altogether. – /iV, uV/ (while retaining /iː, uː/) is already what dictionaries do, isn't it?
 * Unless we want people to be confused? – I don't know how people could possibly be confused. /ˈmiːdiə/ is way more intuitive than /ˈmiːdiːə, -jə/ as far as I'm concerned. Keeping both /iː, uː/ and /i, u/ suggests that /i, u/ can be compressed, i.e. shorter, which is exactly what the lack of length marks suggests. Nardog (talk) 23:42, 23 June 2017 (UTC)
 * There's something wrong with that quote. To me, it looks like Wells wrote it when he had specifically the lax variants in mind and Longman failed to catch that until it was too late.  are definitely phonetically different from . The change from  to  (at least in more conservative RP) has no all the way quality about it (or however you want to put it),  are  transcribed with different symbols. That's the whole difference. Here's the corresponding quote from my CD version of the 3rd edition of LPD:
 * Sometimes a sequence of sounds has two possible pronunciations: either as two separate syllables, or compressed into a single syllable. Possible compressions are shown in this dictionary by the symbol between the syllables affected.
 * lenient Two pronunciations are possible: a slower one, and a faster one . (...)
 * 3. When a syllable is compressed, one of the following phonetic changes takes place. (...)
 * • A weak vowel or  is changed into the corresponding semivowel,  or , producing in combination with the following vowel a crescendo DIPHTHONG.
 * influence (= or ). (...)
 * It makes much more sense. But, if your typical pronunciation of the non-phonemes is lax, then it's possible that you'd distinguish their non-syllabic variants from  as . Again, I'd like to see some evidence of that pronunciation in contemporary RP and GA.
 * I find it strange that you think that my could mean a full vowel . I clearly distinguished between the two. Restoring the length marks, I meant  for RP and  (yes, with a tense ) for GA. Those are the transcriptions that the LPD/CEPD would have to give if they wanted to accurately transcribe the recordings without using the non-phonemes.
 * Why? Because we'd be using them differently from the way most dictionaries use them. How would you moderate hundreds if not thousands of pronunciations copied word-for-word from dictionaries that would use in the word-final position? It's just asking for trouble and inconsistency. Let's either follow many dictionaries that distinguish  from  in their transcriptions or let's just stop distinguishing them as some other dictionaries have already done.
 * /ˈmiːdiə/ is way more intuitive than /ˈmiːdiːə, -jə/ as far as I'm concerned. - I beg to differ. How many users would know that the second can be turned into an approximant?  in happy can never be turned into an approximant. And how is  not intuitive? Mr KEBAB (talk) 01:38, 24 June 2017 (UTC)
 * (Post moved from below) As for the contrasts (qualitative, without phonemic length difference), they are not that uncommon in Europe. Both occur in Low German and Low German-influenced Standard German, some Swiss German dialects and some Northern Dutch dialects. Polish, Ukrainian and Standard Dutch have the  contrast and so does Standard Danish, which actually contrasts long and short  (for convenience, they are most often transcribed ). Mr KEBAB (talk) 13:01, 24 June 2017 (UTC)
 * Are you seeing what I'm seeing? The longer we continue this conversation the more I'm confused. I don't know why you keep writing the exact same notation twice for LPD/CEPD's media examples. And I wrote [ˈmiːdiə], not [ˈmiːdiːə], by which I meant the same thing as [ˈmiːdi̯ə]. Which, obviously, I suppose is what you mean by /ˈmidjə/ since you consider /j/ can only be [i̯].
 * Why would it be inconsistent? By definition /i/ means a high front vowel that could be either shorter or laxer than /iː/, which is exactly what the 'i' in media is. Do you mean that it'd be inconsistent if we used /iː/ for HAPPY yet kept prevocalic /i/? If so, then I'd say why switch HAPPY /i/ to /iː/ in the first place? Isn't the current way already following "many dictionaries that distinguish /i, u/ from /iː, uː/"?
 * How many users would know that the second can be turned into an approximant? – They don't have to. Wikipedia is not a dictionary, they don't have to know every single possible realization of a word. And if /ˈmiːdiə/ and /ˈmiːdiːə, -jə/ mean the same thing, why pick the one that takes more space? You seem fixated on technicality, but the current system is already less ambiguous and more efficient than what you're proposing. Nardog (talk) 04:17, 24 June 2017 (UTC)
 * I did it in the original message to clearly show that recordings in both the LPD and the CEPD have exactly the same phonemes and then just reposted it without thinking about it. Since the transcriptions are exactly the same, what is the problem with that?
 * You wrote when you wrote LPD and CEPD say /ˈmidjə/, you meant something like [ˈmiːdiə] or [ˈmidĭə]?, which I found confusing. You used phonetic brackets instead of phonemic slashes, and wrote " or ", which could (and was) interpreted by me as "did you mean to transcribe a vowel or an approximant?" And, as I said, I found it strange because I clearly differentiated between the two. That is also why I interpreted your first as representing specifically the FLEECE vowel, you also forgot to transcribe the length in your second transcription. I don't mean to "attack back" (I don't perceive your post that way anyway), I'm just explaining what happened.
 * Do you mean that it'd be inconsistent if we used /iː/ for HAPPY yet kept prevocalic /i/? Yes, that's it.
 * If so, then I'd say why switch HAPPY /i/ to /iː/ in the first place? To make our transcriptions more exclusively phonemic. Lax HAPPY is very close to becoming non-RP and non-GA, and it's already non-standard in Australia and New Zealand. I just don't think that we need that symbol and the LPD/CEPD should also stop using it.
 * Isn't the current way already following "many dictionaries that distinguish /i, u/ from /iː, uː/"? Of course. I didn't say it wasn't. But we don't actually seem to use and especially  to represent GOOSE/FLEECE with possible compression. See California, Australia and other entries with funny/wrong transcriptions. If we want to keep using, we must introduce the signs   and  (per the LPD) in order not to confuse people. We also need to write way better notes for them in this article. As of now, they don't mention the phenomenon of compression in a single word.
 * They don't have to. Wikipedia is not a dictionary, they don't have to know every single possible realization of a word. They absolutely do have to know that if they're going to transcribe English words themselves (which, as you know, anyone can do on Wikipedia). Again, see California, Australia, etc.
 * And if /ˈmiːdiə/ and /ˈmiːdiːə, -jə/ mean the same thing, why pick the one that takes more space? Again, to have a more exclusively phonemic transcription.
 * the current system is already less ambiguous and more efficient than what you're proposing. How is using the non-phonemic symbols less ambiguous than using the actual phonemic symbols ?!  are equally as good as  if you understand what they mean. If you're just a newbie that goes "oh look,, just like the ones in my native language", they're obviously worse and more ambiguous than . This is why we need to introduce  and  if we're going to keep the non-phonemes . As long as we do that, I think I can live with these strange (and IMO unnecessary) symbols. Mr KEBAB (talk) 12:56, 24 June 2017 (UTC)
 * Yeah, then there's also Mach's post where he explains that could be interpreted as phonemes in a rather complicated polysystemic analysis. I forgot about that, so bear that in mind when reading the post above. Mr KEBAB (talk) 13:11, 24 June 2017 (UTC)
 * Ahh, now I get why you thought I'd thought "/j/ could mean a full vowel [iː]". I didn't forget the length mark. By "[ˈmiːdiə] or [ˈmidĭə]", I meant "[ˈmiːdiə], which can be alternatively transcribed as [ˈmidĭə]". So it was a yes–no question, not a A-or-B question. But looking back at it now, I can totally see why someone would interpret it the way you did. Pardon my use of words (and symbols).
 * Readers don't necessarily have to know /iV/ means "/iːV/ or /jV/" in order to grasp the pronunciation because in most cases the compression is automatic and/or hardly noticed, and even if they saw /iːV, -jV/ most of them wouldn't be able to tell the difference. As far as editors go, if there isn't enough awareness among them that /iV/ and /iːV, -jV/ are identical, that doesn't automatically mean we should deprecate prevocalic /i/, it just means we can raise the awareness by mentioning compression on this help page and shortening those unnecessarily long transcriptions.
 * /iV/ is less ambiguous than just either /iːV/ or /jV/. Or, would you actually go and replace all existing instances of /iV/ that are subject to compression with /iːV, -jV/? And even if you did, do you think other editors are going to follow suit every time they add a transcription that involves such a sequence? That's precisely "asking for trouble and inconsistency". The fewer alternative pronunciations the less room for inconsistency, and of course more efficiency.
 * There certainly appears to be a prevocalic short high front vowel that is neither typical /iː/ nor /ɪ/, yet not typical /j/ either because of phonotactic reasons. I don't know if that automatically makes it a phoneme, but at least it seems to me it's worth giving a diaphoneme.
 * As for the undertie, I don't know if we need to introduce it right now because the predisposition to compression is already indicated by the lack of length mark in my opinion. Maybe we will need it if we were to switch the HAPPY vowel to /iː/, which, however, I don't think it would be unreasonable for us to keep on hold until a multiple of major dictionaries discontinue transcribing the HAPPY vowel with /ɪ/ for RP. Nardog (talk) 14:52, 28 June 2017 (UTC)
 * Readers don't necessarily have to know /iV/ means "/iːV/ or /jV/" in order to grasp the pronunciation because in most cases the compression is automatic and/or hardly noticed, and even if they saw /iːV, -jV/ most of them wouldn't be able to tell the difference. There's also a large group of non-native speakers that aren't necessarily aware of the automaticity of compression. Plus, even native speakers don't necessarily understand the difference between and . I strongly disagree with the last part of your sentence and I'm not sure how you reached that conclusion.


 * There certainly appears to be a prevocalic short high front vowel that is neither typical /iː/ nor /ɪ/, yet not typical /j/ either because of phonotactic reasons. I don't know if that automatically makes it a phoneme, but at least it seems to me it's worth giving a diaphoneme. It surely is a diaphoneme ("either or ") if we don't want to treat it as a separate unstressed vowel.


 * As for the undertie, I don't know if we need to introduce it right now because the predisposition to compression is already indicated by the lack of length mark in my opinion. We do, to avoid wrong transcriptions on California, Australia etc. Mr KEBAB (talk) 01:44, 2 July 2017 (UTC)

Here's a list of words with vowels transcribed with the transcription I propose (I hope I listed all of them):


 * Monophthongs:.
 * Weak monophthongs:.
 * Diphthongs:.

There are some problems with it:
 * are more similar to than  are. However, to my eyes at least, they are sufficiently different. I'd like to think of our readers as intelligent and I think they'd easily figure out the difference. We could partially fix this problem by transcribing  as, following most BrE dictionaries (we could also switch our  to ).
 * What is now written (as in sawer), in this transcription would have to be written  and the dot could never be omitted in order not to confuse  with the centering diphthong . That is easily fixable by ditching, which I am all for. We don't need it, just as we don't need.

On the other hand, there are some positive things: Mr KEBAB (talk) 19:48, 23 June 2017 (UTC)
 * The phonemicity of the vowel length varies considerably from dialect to dialect. By not using the length marks we don't look like we're making a statement on whether we think that vowel length in English is inherently phonemic or not (again, it's variable, period).
 * The current (mostly conservative RP-based) system does not account for the fact that most speakers of modern RP distinguish from  and  from  mostly by length, not quality. In closed syllables at least,  was always a long(ish)  in RP, at least according to my experience. By not transcribing 'length' we're kind of bypassing that dilemma.
 * Whenever there's a long list of transcriptions, not having the length marks shortens it to a certain extent.
 * I'm still not sure whether we need that level of detail here. We don't really aim at an in-depth phonological analysis of English words. – Of course we don’t, but neither do other dictionaries, yet they still make the distinction.
 * Our transcription system should not aim at being as short as possible, but as helpful to the readers as possible. There are dictionaries that transcribe the word *bit* as “bit”. Of course, they are often not using the IPA, but that does not change my point: The transcription “bit” is inherently ambiguous. By adding a little redundancy and using Gimson’s /bɪt/ vs. /biːt/, we are reducing ambiguity and helping the readers.
 * By not using the length marks we don't look like we're making a statement on whether we think that vowel length in English is inherently phonemic or not. – Yes, we do. The whole point of redundantly transcribing both the length marks AND the different qualities in pairs like /biːt – bɪt/ is avoiding a statement on whether or not vowel length is inherently phonemic. If you drop the length mark, your statment is that length is *not* phonemic. If you drop the quality distinction, then your statement is that length *is* phonemic.
 * It is true that our current Gimson-style system is largely based on British tradition. The system you are proposing is largely based on American tradition (though American dictionaries often use their own idiosyncratic respelling systems). I really think we should keep our current system because of the two advantages I have mentioned: It adds extra-redundancey to make things extra-clear, and it does not make a statement on whether or not we think vowel length is inherently phonemic.
 * I am in favour of dropping /ᵻ/ and /ᵿ/ because they are not much used outside of Wikipedia and I think our Core content policies (apply and) oblige us not to give undue weight to an unusual transcription. However, I think that before throwing them out we should reach out to users who have in the past voiced their support of a “diaphonemic” approach. --mach &#x1f648;&#x1f649;&#x1f64a; 21:12, 23 June 2017 (UTC)
 * I think Mach has expressed my sentiments in a far better way than I could ever have. When we write /ɜːr/, we're not declaring that the length is phonemic. We are merely giving readers a choice and they can interpret it as whatever is equivalent in their own accent, like /ɜr/, /ɜː/, or /əː/. If we were to drop the length simply because it is not "inherently phonemic", the same argument could be made for dropping word-final or preconsonantal /r/. Nardog (talk) 22:09, 23 June 2017 (UTC)
 * Random observations:
 * In your proposal caught would be /kɔt/, which is how cot is transcribed in some accents. The length mark helps remove this ambiguity in accents that retain the cot–caught distinction (in part, if not in large part, by length) such as RP and Australian.
 * /e/ is how /eɪ/ is often transcribed in the US. Probably not a good idea.
 * I'd like to think of our readers as intelligent and I think they'd easily figure out the difference. – Well that is a bold thing to think of them as. If they were, we would have deprecated Respell years ago.
 * Seriously though, if speakers of any varieties of English (native and ESL alike) were asked what the difference of bit and beat is, I'm 100% certain the vast majority would say "length". And if you asked someone who speaks English and has never seen IPA to tell /biːt/ and /bɪt/ apart, I'm sure they would be able to answer correctly, while if they were shown /bit/ and /bɪt/, the answers would be 50/50 (or 60/40 or something, because $\langleɪ\rangle$ is visually shorter than $\langlei\rangle$).
 * Also, most readers unfamiliar with IPA and briefly visiting Wikipedia are not going to be given even a chance to actually compare $\langlei, u\rangle$ and $\langleɪ, ʊ\rangle$ and "figure out the difference", until they encounter a transcription that contains both or see enough articles to notice there even being a difference. Ideally, whether a symbol stands for FLEECE or KIT (or FOOT or GOOSE) should be clear without seeing the symbol for the other.
 * If we could attain visual intuitiveness to linguists and laypeople alike while not compromising the scientific precision (as it's "equally 'scientific'" according to Wells), I'd say that's killing two birds with one stone. Nardog (talk) 14:19, 28 June 2017 (UTC)
 * Ok, this all sounds pretty reasonable.


 * Yes, we do. The whole point of redundantly transcribing both the length marks AND the different qualities in pairs like is avoiding a statement on whether or not vowel length is inherently phonemic. That's true, but generally when you use the length mark in phonemic transcription you're actually making a statement that vowel length is phonemic. So it could be seen either way. Again, there's no perfect solution.


 * Seriously though, if speakers of any varieties of English (native and ESL alike) were asked what the difference of bit and beat is, I'm 100% certain the vast majority would say "length". And if you asked someone who speaks English and has never seen IPA to tell /biːt/ and /bɪt/ apart, I'm sure they would be able to answer correctly, while if they were shown /bit/ and /bɪt/, the answers would be 50/50 (or 60/40 or something, because $\langleɪ\rangle$ is visually shorter than $\langlei\rangle$). And I'm certain that at least a decent percentage of them would perceive as having both a different quality (however they'd describe it) and a shorter duration than . Especially if they're aware of pre-fortis clipping. Unless they're from Australia or Birmingham, where  for  is widespread. And I think that you're overestimating how often  would be mistaken for the  vowel. That symbol is very rarely used that way. Mr KEBAB (talk) 01:44, 2 July 2017 (UTC)

I'm trying to keep up with this discussion, but I'm not at all confident that I am seeing the right symbols on my screen. The particular problem for me is the Oxford (OED, C. Upton) symbols for "either ə or ɪ" and "either ə or ʊ". In printed material I've seen, the former appears as a small cap I (ɪ) with a bar through it and the latter is ʊ with a bar through it. But in some of the discussion above, and on the Help page itself, the former symbol looks like an i with no dot but with a bar. Does anyone know if this is an intentional difference? RoachPeter (talk) 08:57, 28 June 2017 (UTC)
 * I'm sure $\langle ɪ \rangle$, the symbol the OED uses for "free variation between /ɪ/ and /ə/", is what we mean by "ᵻ". Arial, the default sans-serif font on Windows, for some reason shows the wrong symbol you described when you type "ᵻ". If you specify one of Windows' other pre-installed fonts such as Times New Roman, or manually strike through "ɪ" via HTML, you'll see the intended symbol: ᵻ, ɪ
 * You can see the same symbol at OED's key page. Nardog (talk) 11:21, 28 June 2017 (UTC)

short, closed /u/
The sound of the first vowel of "Ukraine" in the sound files of all dictionaries i checked (and our ) is not the long sound in "true" but not always the short, open sound of "put" either. It's a short but closed /u/. We need to change IPAc-en to accept u i.e. juˈkreɪn (it now produces the error message ), and we need to mention this sound on Help:IPA for English. Now our transcription incorrectly makes non-native speakers believe the pronunciation of words with this sound has a long vowel. --Espoo (talk) 05:38, 2 July 2017 (UTC)
 * The pronunciation with the long is neither wrong nor non-native.  is an unstressed vowel that is considered to be a phoneme only by some scholars.  represents a neutralization between  and  so that you can choose either vowel. According to Geoff Lindsey,  is a superfluous symbol and we should just write  instead, but, AFAIK, that's a minority point of view. You should also remember that unstressed English vowels are shorter than the stressed ones.


 * We should add . Mr KEBAB (talk) 05:52, 2 July 2017 (UTC)


 * Yes, because unstressed English vowels are shorter than the stressed ones, we should not mislead non-native speakers by marking them as long. You also agree that we should add /ju/.
 * On the other hand, you say that the pronunciation of "Ukraine" with the long /uː/ is neither wrong nor non-native although this contradicts what you just said about unstressed syllables. The long pronunciation is very rare in normal speech, especially in a sentence. It probably only occurs when the word is said in a quite unnatural way separately and in a self-conscious way, when the term is still new or by people who rarely use the term. --Espoo (talk) 08:29, 2 July 2017 (UTC)


 * First, the type of transcription that we use is phonemic, not phonetic (allophonic), therefore by definition it can't cover shortening of the 'long vowels' in unstressed positions. Also, you're being deceived by the length mark. The presence of it doesn't mean that the vowel has the same length in all environments (again, we use phonemic transcription) or even that it's always long. For instance, the in readership is quite short due to rhythmic clipping. In Tarkington, the  is even shorter due to both rhythmic clipping and pre-fortis clipping. The same applies to unstressed vowels, which are also shorter than their stressed counterparts in the same environment.


 * As I said, the symbol $\langle\rangle$ represents a neutralization between and  and sometimes also . Only some scholars consider it to be a separate phoneme. Mr KEBAB (talk) 13:56, 2 July 2017 (UTC)

RfC: proposing (in part) to remove IPA from the lead
There is currently an RfC proposing removing many things from parentheticals in the lead, including IPA. Please join in the discussion at Wikipedia talk:Manual of Style/Lead section. Curly "JFC" Turkey 🍁 ¡gobble! 06:18, 3 July 2017 (UTC)

Move discussion in progress
There is a move discussion in progress on Help talk:IPA which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 16:16, 15 July 2017 (UTC)

/aɪ/ versus /ʌɪ/
According to the cite note for /aɪ/, "dictionaries do not generally record [/ʌɪ/], so it has not been allocated a separate transcription here." Oxford seems to record it, though. Is it worth having another look at this? Rovingrobert (talk) 08:54, 29 June 2017 (UTC)
 * Some dictionaries use $\langle\rangle$ instead of $\langle\rangle$, not alongside it. This usage has been criticised by e.g. John Wells as phonetically inaccurate. Mr KEBAB (talk) 14:22, 29 June 2017 (UTC)
 * I found this article (see Section 8) thorough and to the point on this matter. Nardog (talk) 11:02, 21 July 2017 (UTC)

Iruhiqjeraid
Upon noticing that there was an attempt to move the main page to encompass English and Iruhiqjeraid, what is the Iruhiqjeraid language and is/was its pronunciation similar to that of English? Rovingrobert (talk) 13:46, 13 June 2017 (UTC)
 * It was a troll trying to be noticed. There's nothing more to it. Mr KEBAB (talk) 13:52, 13 June 2017 (UTC)
 * However, it would be nice to have that notice removed. Maybe someone can take it to the admins. Mr KEBAB (talk) 03:33, 15 June 2017 (UTC)
 * At least it was a fairly clever, multi-website troll. 110.142.225.47 (talk) 12:23, 15 September 2017 (UTC)
 * Clever? No, the opposite. There are much more smart ways to troll, which I obviously won't discuss here (see WP:TEMPTATION). Anyway, the issue was fixed by moving Help:IPA for English to the current name. Mr KEBAB (talk) 15:06, 15 September 2017 (UTC)

Nasal vowels
Right now the table lists only one nasal vowel /ɒ̃/ (seems to be a merger of two French vowels /ɑ̃/ and /ɔ̃/). However, I've been trying to conform it with the CEPD and add three additional nasal vowels; I've spent quite a time on it and finished it already, but suddenly I've become hesitant to send the changes. I mean four vowels: /æ̃, ɑ̃ː, ɔ̃ː, ɜ̃ː/ (French /ɛ̃, ɑ̃, ɔ̃, œ̃/ respectively). Is there a need for this?--Lüboslóv Yęzýkin (talk) 19:29, 21 June 2017 (UTC)
 * I don't have an exact ref right now, but AFAIK Wells says that most Brits are unable to distinguish from . I'd be hesitant to transcribe both. Mr KEBAB (talk) 19:46, 21 June 2017 (UTC)
 * I was actually going to bring this up. OED recognizes /æ̃/ in addition to /ɒ̃/, contrasting fin and bon. American dictionaries such as AHD tend to indiscriminately transcribe French vowels (by adding superscript "N" to their English vowel symbols), so I don't think they're much of use as reference. /œ̃/ is merged with /ɛ̃/ in Parisian French, so I'd be against adding it (or its English counterpart). Since our system is phonemic, what we should take into account is how many vowels are contrasted, rather than English-speaking realization of French vowels. But that obviously could be tricky for marginal segments.
 * I think I've read somewhere that unrounded vowels of French in general or at least nasal ones still tend to be more rounded than those of English, and /ɑ̃/ and /ɔ̃/ are indeed very hard to distinguish to non-native ears, so I'll echo Mr KEBAB in /ɒ̃/ accounting for both /ɑ̃/ and /ɔ̃/. But I think there's a room for /æ̃/, i.e. un bon vin blanc /æ̃ bɒ̃ væ̃ blɒ̃/. Nardog (talk) 20:35, 21 June 2017 (UTC)
 * I think we should go with what native speakers of English do. If the majority preserves the distinction between and, then we should say that. In German and Belgian Dutch, the vast majority of speakers that preserve the nasals would say , with a rounded . The pronunciation  would just sound off, even though it is standard in Paris. Mr KEBAB (talk) 21:29, 21 June 2017 (UTC)
 * You're right, it would be nice if we could find a source as to how many nasal vowels in French loans are distinguished by English speakers (though unlikely, because eventually it comes down to how acquainted with French they are). That CEPD paragraph is simply saying that they are transcribing the French vowels using approximate English phonemes (similar to AHD) and not claiming that those are contrastive, so it's not really a useful source either. What I can see being added to the table (and to IPAc-en) so far is just /æ̃/ (per OED).
 * If one needs to transcribe a word distinguishing all four of them, that means it's probably not completely assimilated into English, and at that point one should just use IPA-fr. Nardog (talk) 22:22, 21 June 2017 (UTC)
 * Is this the Wells thing you were talking about? He says in "BrE-accented French" (not sure if he means when they try to speak French words or French sentences) /œ̃/ is nasalized /ʌ/ (interestingly, not round). This is probably influenced by the fact that /œ̃/ is usually written "un".
 * I also wanted to find out OED transcribes what is /œ̃/ in French with /æ̃/ or /ɒ̃/, but I couldn't think of an example to look up. Nardog (talk) 22:44, 21 June 2017 (UTC)
 * Maybe, but the point is that it says exactly what I remembered. Thanks. Mr KEBAB (talk) 22:59, 21 June 2017 (UTC)
 * is pretty rare in French (aside from in the indefinite article), so maybe there are no halfway-adapted loanwords that have it. The vowel list in the OED's pronunciation key doesn't list an equivalent for it. — Eru·tuon 00:00, 22 June 2017 (UTC)
 * LPD has one (really, one) word with it: vingt-et-un, which is a card game. The French pronunciation is, the RP pronunciation is and the GA pronunciation is . According to LPD's key to phonetic symbols,  has no equivalent in GA. I'd say let's add it, not least because the variant with  is the preferred one. Mr KEBAB (talk) 01:33, 22 June 2017 (UTC)
 * The problem is: how would we transcribe it? is slightly inappropriate, since RP is a non-rhotic accent. Then again,  would be inconsistent with other vowels. I think  is the best symbol. Mr KEBAB (talk) 01:48, 22 June 2017 (UTC)
 * That seems misleading to me, as an American English speaker: I would not pronounce the un part of vingt-et-un with an r-vowel (vant-ay-urn). I suppose it wouldn't be misleading, technically, if it were prefixed by "RP only" or "non-rhotic varieties only". On the one hand, it doesn't matter that much because it will hardly ever be used. On the other, why add it? Is there an article it would be used in? And if it's diaphonemic, what dialects is it intended to represent? Do Southern Hemisphere Englishes use a phoneme corresponding to it, or is it just RP? If it's just RP, it's not really diaphonemic. Confusing situation. — Eru·tuon 03:17, 22 June 2017 (UTC)
 * We could add it so that we can have a complete list of nasal vowels. Then again, mentioning it in a footnote seems to be enough to me. As for the Southern Hemisphere thing, I have absolutely no idea. Preserving in that word seems to me to be most likely in South Africa, since their Cultivated speakers seem to be a bit obsessed about RP. Mr KEBAB (talk) 03:27, 22 June 2017 (UTC)
 * $\langleɜ̃ːr\rangle$ looks incredibly silly to me, even though it just so happens to be a pretty probable representation of how the vowel is pronounced in Quebec! Just like CEPD, LPD's $\langleɜ̃ː\rangle$ seems to be the result of automatically allocating the French vowels to approximate English phonemes, as opposed to how they are actually produced by English speakers, and just one entry from one dictionary isn't too strong of evidence to add it. Nardog (talk) 03:31, 22 June 2017 (UTC)
 * They specifically say that the variant with is the preferred one as far as RP is concerned. Mr KEBAB (talk) 03:36, 22 June 2017 (UTC)
 * In LPD (1st ed., only one I had access to), Wells writes:
 * Many speakers of BrE have some knowledge of French. Accordingly, they may well use nasalized vowels (or attempts at them) and other characteristics of French pronunciation when they pronounce words they perceive as French. In fact the sounds represented as ɒ̃, æ̃, ɜ̃ː (as in bon, vingt-et-un) may for this reason regarded as marginal members of the RP system. On the whole, though, it is only those whose knowledge of French goes beyond the average who succeed in differentiating the real-French vowels ɔ̃, ɑ̃ (as in son and cent respectively) instead of confusing them as ɒ̃.
 * And CEPD uses /ɜ̃ː/ even for US pronunciation of vingt-et-un. So I guess I won't mind adding /ɜ̃ː/, but not /ɜ̃ːr/ or /ɑ̃ː, ɔ̃ː/.
 * (While I'm at it, I wonder why they (LPD & CEPD) chose /ɜ̃ː/ to transcribe French /œ̃/. In the post Wells says it's usually nasalized /ʌ/, and /œ̃/ looks fairly close to RP /ʌ/ at least according to French phonology. But thinking further, maybe /œ̃/ does kinda sound like RP /ɜː/, possibly because of the length. Dunno.) Nardog (talk) 11:57, 22 June 2017 (UTC)
 * Well, I have only the CD of the 3rd edition of LPD but not the actual physical book. My local library sucks, so I don't think they have it either. I could ask around, but I can't promise anything. Thanks for the quote, I agree with Wells (and you).


 * Most likely because is a free vowel, whereas  is checked. Then again,  is also checked, and that doesn't stop LPD and CEPD from transcribing French  as  (which is phonetically correct for Parisian (but not Quebec) French by the way). Dunno either. Mr KEBAB (talk) 12:40, 22 June 2017 (UTC)

Thanks for the discussion. Now I tend to think that I simply wasted my time. Though, I think that there is still a room for /æ̃/.

What is certain that I'm now strongly against any substitution for /œ̃/. Actually exactly this sound made me doubt for a moment and prevented me from sending the changes. As you can see, even the CEPD could not have come up with any example of an English word with this sound other than the proper name Lebrun which, I believe, most English speakers would pronounce with /-ʌn/. Of course, we can find other proper names like Verdun which, again, most speakers would pronounce in a truly English way, /vɛərˈdʌn/; and I doubt any French name should be transcribed with an Anglicized pseudo-French pronunciation which may resemble the French one, but yet it remains corrupted. Either it is fully Anglicized, or we must just provide the correct French transcription, and then anybody can try one's best to pronounce it. Bearing in mind, that in French itself /œ̃/ is rare, it is even surprising that English has even ended up with one loanword with it (vingt-et-un). Yet still it is pronounced with /-ˈəːn/, as stated by the ODE. There is yet another loanword, eau de parfum, though the CEPD suddenly gives /-ʌm/.

I also agree about the merger of /ɑ̃/ and /ɔ̃/ in English. Actually in my change I had written a note which would state these two vowels are in free variation. What I'm still unsure of is how to write this vowel. I have a suspicion that /ɒ̃/ in the ODE is simply an ad hoc symbol rather than how people may actually pronounce it. Why do they use the nasalized checked vowel of LOT for it? Why does the CEPD instead use free /ɔ̃ː/ (with the length mark)? A counter argument, however, may be both the dictionaries use checked /æ̃/.--Lüboslóv Yęzýkin (talk) 19:20, 22 June 2017 (UTC)
 * I've just checked LPD, and it does have an entry for parfum. The preferred variant for RP is..., with a short that AFAICS is not used in any other entry. Mr KEBAB (talk) 19:59, 22 June 2017 (UTC)
 * I don't think anybody here has expressed reluctance about adding /æ̃/. That one seems uncontroversial. Nardog (talk) 20:56, 22 June 2017 (UTC)
 * AFAICS we have a clear consensus (perhaps among too few users though) to add both and  (which, according to the LPD, is  in RP and  in GA). Mr KEBAB (talk) 21:08, 22 June 2017 (UTC)
 * Found OED transcribes falun(s), "a French provincial term for the shelly tertiary strata of Touraine and the Loire" (what?!), as /ˈfælæ̃/ (conformed to our system). So apparently /œ̃/ is indeed represented by /æ̃/ (/ã/) in OED's system. Nardog (talk) 23:05, 22 June 2017 (UTC)
 * On a side note, /ɒ̃/ and /æ̃/ may not only be found in French calques; the equivalent phonemes are also native to Polish. Rovingrobert (talk) 09:37, 29 June 2017 (UTC)

✅ Since there's no harm in adding a diaphoneme, I've implemented /æ̃/. In fact there already were articles like Poincaré conjecture and Kalki Koechlin, where /æ̃/ would have been more appropriate than /ɒ̃/. Nardog (talk) 02:10, 24 June 2017 (UTC)
 * Also Lech Wałęsa. 110.142.225.47 (talk) 12:19, 15 September 2017 (UTC)
 * I didn't realize native speakers of English attempt to pronounce Polish nasal vowels. I've always thought that the most logical anglicization of Wałęsa is . Mr KEBAB (talk) 14:45, 15 September 2017 (UTC)
 * Well, it's about as plausible as pronouncing the French equivalent. /æ̃/ is sometimes lenited to /æn/ in names like Baudin, Rodin, etc. Rovingrobert (talk) 14:41, 17 September 2017 (UTC)
 * The thing is that at least in the past, English (especially RP) speakers were far more likely to attempt to pronounce French loanwords as closely to original versions as possible than to do the same with Polish. Even now many people just don't feel like researching how something is pronounced before saying it (yes journalists/commentators, I'm also talking about you). Plus, Polish nasal vowels have severe phonotactic restrictions, because they can appear only before fricatives, a pause, never word-initially and, in the case of, often also not word-finally. All of this makes it unlikely that someone would use or  in Polish loans. If you do, you're probably an exception to the rule. Mr KEBAB (talk) 16:03, 17 September 2017 (UTC)
 * True, but not transcribing Polish nasal vowels as we do French means we're making a judgement call as to how they should be adapted into English. Rovingrobert (talk) 06:19, 18 September 2017 (UTC)
 * But is there any article in which we proscribe an anglicization of a Polish name? In Lech Wałęsa for example, all we're doing is reporting on how native speakers of English pronounce his name according to reputable sources. If there isn't an established anglicization of a Polish surname, we shouldn't provide English IPA for it. It's as simple as that. Mr KEBAB (talk) 14:18, 18 September 2017 (UTC)

How about we change /ɔər/ to /oʊr/?
It recently occurred to me and I wonder, how many would oppose if we proposed that /ɔər/ be changed to $\langleoʊr\rangle$? Our current notation $\langleɔər\rangle$ is rarely attested, and, although there is also a case to be made that we should merge /ɔər/ with /ɔːr/ because many dialects have, it seems to me it should at least be changed to $\langleoʊr\rangle$ even if we decide to keep it as a distinct diaphoneme.

Notation in dictionaries

Currently all of the major dictionaries freely accessible online that still include pronunciation variants with unmerged FORCE, which are few nonetheless, employ /oʊr/:
 * Cambridge Academic Content Dictionary, Cambridge University Press, dictionary.cambridge.org
 * Dictionary.com Unabridged, Random House, dictionary.com
 * Webster’s New World College Dictionary (4th ed.), Houghton Mifflin Harcourt, 2010, collins.com
 * Random House Kernerman Webster's College Dictionary, K Dictionaries, 2010, thefreedictionary.com

In pronunciation dictionaries, Longman Pronunciation Dictionary (3rd ed., 2008) still includes the variants, again as /oʊr/. English Pronouncing Dictionary has dropped /ɔə/ long since, and I don't know if it ever included the American counterpart since its 15th ed. (1997) added American pronunciations, but 16th ed. (2003) apparently didn't so unlikely. The Routledge Dictionary of Pronunciation for Current English (2017) doesn't include the variants; I don't know about its predecessor, The Oxford DPCE (2001), but OED3, also supervised by Clive Upton, doesn't, so I'm guessing it didn't either. A Pronouncing Dictionary of American English had /or/ for the unmerged FORCE, which is actually in line with the other dictionaries using /oʊr/ because it had /o/ for GOAT.

FWIW, The American Heritage Dictionary of the English Language had included variant pronunciations with "ōr", i.e. /oʊr/, in its 4th ed. (2000), but they were dropped in 5th ed. (2016). I don't know when Merriam-Webster (which is usually pretty liberal w/r/t pronunciation) stopped transcribing them (or if it ever did), but its pronunciation guide states: "the number of speakers that make such a distinction is currently very small, and we have not represented the distinction in this dictionary."

The trend to drop the distinction is likely to continue on both dictionaries' and speakers' part, and so it's becoming increasingly harder to verify with a reliable source whether a word has /ɔər/ or /ɔːr/. So there is a case to be made that we should drop /ɔər/ as a diaphoneme altogether.

Problems with $\langleɔər\rangle$

The biggest problem with $\langleɔər\rangle$ is obviously that the vowel is no longer pronounced (nor transcribed) like that in dialects that distinguish it from NORTH. But there's more.

All the other diaphonemes for the (historically) pre-R vowels (/ɑːr, aɪər, aʊər, ɛər, ɪər, ɔːr, ʊər/) are more or less consistent in that if you take out the Rs they will perfectly resemble the RP phonemes (/ɑː, aɪə, aʊə, ɛə, ɪə, ɔː, ʊə/), and if you take out the length or schwa they will perfectly resemble the GenAm realizations (/ɑr, aɪr, aʊr, ɛr, ɪr, ɔr, ʊr/). But FORCE is no longer pronounced as [ɔə] in RP, and if you take out the schwa it won't be distinguishable from /ɔːr/. In some parts of the US it is still pronounced like [ɔə], but so is NORTH, again rendering the current notation ineffectual in terms of letting the reader identify what phoneme it represents.

Now I realize this argument can also be made against $\langleoʊr\rangle$, but $\langleoʊr\rangle$ still has at least one more advantage $\langleɔər\rangle$ doesn't: It doesn't require readers to remember one extra phoneme. And since the unmerged (rhotic) FORCE is phonemically /oʊ/ + /r/, $\langleoʊr\rangle$ only makes more sense as an option. Although a bit of an outlier as far as the rhotic vowels go, $\langleoʊr\rangle$ does at least allow the speakers of non-rhotic accents to think of what it would be if a vowel ensued, unlike $\langleɔər\rangle$.

Thus I think /ɔər/ should at least be notated with $\langleoʊr\rangle$ even if we keep it as a diaphoneme distinct from /ɔːr/. Apologies for the rambling, just throwing ideas out there. Nardog (talk) 10:09, 19 September 2017 (UTC)
 * I think that $\langleoʊr\rangle$ ought to be used instead of /ɔər/ because /oʊ/ is currently a more accurate representation of the HOARSE vowel in dialects that have the horse-hoarse distinction. Tharthan (talk) 16:54, 19 September 2017 (UTC)
 * In my opinion, the choice between and  is arbitrary. However, we do write, so  is a somewhat better choice. Furthermore, our IPA is almost fully based on the traditional transcription of RP (which writes ), with the exception of  and , which are written  in that system (the same applies to diphthongs). I think we better keep transcribing it with , but this isn't a very huge issue to me. Mr KEBAB (talk) 14:59, 21 September 2017 (UTC)

Should we explicitly recommend /ɪ/, /ʊ/, etc. over /ə/?
Now that we're deprecating /ᵻ, ᵿ/, what should this help page recommend editors do when transcribing a word with unstressed /ɪ, ʊ/ that can be /ə/ in certain dialects? Is it a good idea for this page to explicitly state that we prefer unstressed /ɪ, ʊ/ (or any unstressed vowel subject to reduction as a matter of fact) over /ə/ and over both (e.g. ), since notating both is is inherently redundant in our polylectal system (the reason we deprecated /ᵻ, ᵿ/) and a more specific vowel is less ambiguous than /ə/? I'm of the opinion that the shorter the notation the better (and so are many, apparently), and this way we could shorten transcriptions like too, which were produced when we deprecated /ɵ/. This is in line with what we already do: preferring more specific segments even if they're merged with others in some accents. Nardog (talk) 00:25, 3 August 2017 (UTC)
 * I'm leaning towards /ɪ/, /ʊ/, with the assumption that those dialects that reduce these to /ə/ do so in a predictable fashion. Am I correct in this assumption? — Æµ§œš¹  [lɛts b̥iː pʰəˈlaɪˀt] 05:38, 3 August 2017 (UTC)
 * I favour flexibility and redundancy. There is no point in prohibiting variants. It would estrange people who, for instance, identify with their native place and care about its pronunciation. The “apparently” you mentioned shows people who are opposed to the IPA per se – the question was not at all about pronunciation variants, and it also seems that a majority would not change the status quo. --mach &#x1f648;&#x1f649;&#x1f64a; 06:09, 3 August 2017 (UTC)
 * Your statement about those who "identify with their native place and care about its pronunciation" strikes me as odd, because that is exactly what this page is already proscribing in the last paragraph of § Dialect variation. If we allowed editors some leeway in transcriptions like you say, then how much should we allow? Flexibility could lead to a creep of alternative pronunciations in lead sections, which is what led some people to advocate for getting rid of parenthetical pronunciations in the RfC (granted, they can be moved to footnotes notwithstanding) . So it's not just redundancy that I'm worried about; rather, I'm worried about the potential inconsistency that being flexible would bring in.
 * A possible compromise solution to this is to encourage applying WP:ENGVAR where applicable to some extent (which I can actually see an argument for in words in the BATH and CLOTH sets), but then again, to what extent? Should we accept transcriptions using IPAc-en like /ˈɒksfəd/, /təˈrɒnoʊ/, and /əˈstreɪliə/, which, according to WP:PRON, must be replaced either with /-ər-/, /-nt-/, and /ɒ-/ or with non-phonemic transcriptions linked to the appropriate article other than this help page? Or is it just vowels? What about rhoticity, flapping, yod-dropping, and so on?
 * (For the record, there is IPA-endia, but it's scarcely used, and International Phonetic Alphabet chart for English dialects it links to is way too narrow for anyone who has little (or even some) acquaintance with the IPA. Maybe this Help page (or a subpage or something) can have a semi-phonemic-semi-allophonic chart for different dialects corresponding to the diaphonemes, in a similar fashion to the tables at Wiktionary, OED, or the way the IPA charts for English dialects article used to be. Alternatively, we may modify or split that article to accommodate both phonemic and narrow descriptions.) Nardog (talk) 01:37, 4 August 2017 (UTC)
 * The problem is that there are words that can be pronounced with RP with either or  and words in which only  is an RP variant. Biscuit is one of them. We should explicitly label both variants as RP and GA (rather than UK and US, in the case of the former abbreviation it's way too ambiguous) in order not to hide information from our readers. Mr KEBAB (talk) 10:48, 3 August 2017 (UTC)
 * Many existing IPAc-en transcriptions are unsourced and/or of proper names, and it is only words that dictionaries have entries for that one can find sources for in both RP and GA in the vast majority of cases. So I guess I won't oppose notating both, especially if both pronunciations can be heard in one dialect and the notation is specifically about that dialect. But like I said above in response to Mach, so long as you're using IPAc-en and linking the notation to this page, you're sort of making a statement that the notation is panlectal or at least polylectal, and therefore notating both unstressed /ɪ/ etc. and /ə/ is inherently redundant. (So maybe we should remove all variation labels from IPAc-en.) Nardog (talk) 01:37, 4 August 2017 (UTC)
 * But what about readers that want to learn RP or GA (no matter whether they're native speakers or not)? Should we force them to check sources themselves? Weak vowel merger is not something that is overly obvious. There are two variants of in RP: strong and weak. Strong  can appear in both stressed and unstressed syllables, whereas the weak  appear only in unstressed syllables and are prone to centralization (which, of course, is inaudible in the case of  which is central for most speakers of RP). General American features only the strong.


 * Examples of the strong in unstressed syllables include the suffix -ive  and (sometimes) the prefixes in- and im-. I'm not sure how widespread is the strong  in unstressed syllables though. There's the suffix -hood, which has the strong.


 * Also, there are dozens of compound words in which lose their original secondary (or tertiary) stress.


 * We must remember that English WP is not only for native speakers of English, let alone native speakers of RP/GA. Mr KEBAB (talk) 11:28, 26 September 2017 (UTC)

Wild growth
This article is about pronunciation, not about spelling. Regularly editors come in and start adding additional example words with variant spellings of the same phoneme. Most of the time this does not add useful information for the reader and is confusing the issue. Those additional spellings can be (have been already) added in English Orthography. Here we should by preference have only one (the word used for the lexical set) and perhaps in some cases a few phonetic variants. For example for consonants examples are given for initial, intervocalic and final positions. For vowels there is not such a clear categorisation and we should restrict to adding only words that carry specific phonetic diversity.&minus;Woodstone (talk) 09:06, 3 October 2017 (UTC)
 * The words I added have syllable-final lax vowels, which are highly unusual in English. They should stay. Mr KEBAB (talk) 10:18, 3 October 2017 (UTC)
 * How would other editors know the reasons for inclusion of each example word. We should do a thorough clean up and make sure that all retained words represent phonetic differences. Reasons for inclusion should be documented (here or in footnotes). &minus;Woodstone (talk) 12:24, 3 October 2017 (UTC)
 * Are you looking at the history correctly? He removed 19 words, added 5 and did explain the reason for inclusion. I don't know what your problem is. Nardog (talk) 13:44, 3 October 2017 (UTC)


 * I was not talking only at about that specific editor. Over time the number of examples keeps growing. I have removed many of them in the past, but it is a recurring problem. For most of the multiple words I cannot fathom the reason for inclusion. &minus;Woodstone (talk) 14:33, 3 October 2017 (UTC)

Do GA and RP have different stress distributions, or do dictionaries?
I'm not sure about your edits in the footnotes. A minimal pair like untidy–and tidy or unorthodox–an orthodox doesn't say much when, since the schwa never occurs in a stressed syllable, one could always just allocate secondary, tertiary, or quaternary stress to [ʌ] and regard it as a conditioned allophone of one /ə/ phoneme that surfaces whenever stressed.

Wells says in this post:"Some would claim that this is a non-issue, because STRUT is always stressed and schwa is never stressed. This argument might work if we define stress lexically, but it will not hold if by stress we mean a rhythmic beat in running speech."and in another:"Some analysts (particularly Americans) argue in the other direction, claiming that the presence of a strong vowel is sufficient evidence that the syllable in question is stressed. In the British tradition we regard them as unstressed."Notice he's not talking about any specific dialect. So it seems to me the difference between the notations you see in British dictionaries like /ʌn-/ and the ones in American dictionaries like /ˌən-/ stems not from the dialects per se but from the dictionaries' preferred phonemic analyses.

For what it's worth, The Oxford/Routledge Dictionary of Pronunciation for Current English had /ˈbeɪsbɔːl/ for RP and /ˈbeɪsˌbɑl/ for GA, claiming "AM has a heavier stressing pattern than BR ... Thus AM may receive two stress marks when BR exhibits only one", but this view was pretty much outright dismissed by Jack Windsor Lewis.

So it seems to me whether STRUT and commA, or NURSE and lettER, are complementary or contrastive totally depends on how you treat non-tonic full vowels, regardless of dialect. Nardog (talk) 14:03, 3 October 2017 (UTC)
 * As far as I know, secondary stress is something that can be objectively detected on formant values. Have there been any studies on this? A commenter under the second post on Wells's blog you've linked to asked the same question.
 * Another question that I'd ask is the following: are there any sources with transcriptions of AmE that use the secondary stress symbol alongside a separate symbol for the STRUT vowel? If not, there's your answer. It's only in word-final and perhaps sometimes also word-initial (probably not true in the case of RP) positions in which an unstressed can be any opener than a true-mid vowel . If you ask me: yes, the evidence is overwhelmingly phonetic rather than phonemic, especially in RP and Australian English.
 * I don't find the American analysis convincing enough, but that's just me. If anything, we should mention both analyses, unless the answers to my questions are 'no' and 'no'. Mr KEBAB (talk) 16:26, 3 October 2017 (UTC)
 * Well, according to Ladefoged, a stressed syllable is "usually, but not always, on a higher pitch" in declarative utterances (Ladefoged & Johnson 2010:111), so F0 values may not have the final say. Hayes (1995) defines a secondarily stressed syllable as one where the vowel is not tonic or schwa and the onset is not subject to flapping, [t] insertion, /l/ devoicing, or aspiration. Which is a similar approach to Wells' syllabification, except applied phoneme by phoneme, not word by word, so the final syllable in Afghanistan will bear a secondary stress even though the /t/ could not possibly be aspirated.
 * In any case, it is true that unstressed /ʌ/ is seen in GA transcription like in LPD and CEPD, so to say "/ʌ/ occurs only in stressed syllables in General American" wouldn't be accurate. Nardog (talk) 18:58, 4 October 2017 (UTC)
 * I probably thought that we're assuming a full STRUT-COMMA merger in General American. I think we can cover both analyses in the guide. Mr KEBAB (talk) 10:37, 10 October 2017 (UTC)
 * We can, but we also don't want to obscure what is truly helpful to readers and editors (such as the fact that $\langleə\rangle$ in some dictionaries could mean what we mean by /ʌ/) by going into too much detail, because this is not an encyclopedic article after all. The current description does cover both analyses, albeit rather implicitly. (Feel free to boldly edit AFAIC, though.) Nardog (talk) 16:03, 10 October 2017 (UTC)
 * The current version looks decent to me, so I'd rather not touch it. Mr KEBAB (talk) 08:03, 16 October 2017 (UTC)

Semi-protected edit request on 18 October 2017
This is a minor edit, but I wanted to bold the 2nd i on historic inside the ɪ box because the 2nd i also makes the /ɪ/ sound.

This is the section I want to change: ɪ PixelPepper (talk) 22:52, 18 October 2017 (UTC)
 * My bad, we should change the word. The purpose of that example was to illustrate word-internal syllable-final . The first of historic is like that, but not the second one. Mr KEBAB (talk) 22:59, 18 October 2017 (UTC)
 * Wouldn't history work just as well? Nihlus 23:46, 18 October 2017 (UTC)
 * History has a stressed in a closed syllable. We need an unstressed  in an open syllable. Mr KEBAB (talk) 23:55, 18 October 2017 (UTC)
 * Red information icon with gradient background.svg Not done: please establish a consensus for this alteration before using the template.  Sak ura Cart elet   Talk 00:29, 19 October 2017 (UTC)