Help talk:IPA/English/Archive 15

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 10

←

Archive 13

→

Bad

Latest comment: 8 years ago101 comments7 people in discussion

I would just like to reiterate for the record (as I have done elsewhere in the past) that what is being done here is terribly wrong and misleading to large numbers of people. If you're going to use the IPA for pronunciations, use it in the standard way. If you attempt to use it in a non-standard way (inventing your own symbols for diaphonemes and so on - made worse by the use of slashes that make the diaphonemes look like ordinary phonemes), people will not know what's going on - the fact that it IS a known standard means that readers will not click through and work their way through this page (which is largely undecipherable anyway); they will assume that what they see is what they are supposed to get, and go away believing either something totally wrong about the pronunciation of some word, or just that Wikipedia is clueless about English pronunciations. W. P. Uzer (talk) 15:17, 14 April 2015 (UTC)

Since our conversation at WT:MOS/P, we have altered the system to fix the one issue you could point to where a confused user might misread our system in such a way as to produce a pronunciation different than what we intend to convey. As such, without any further hypothetical examples, I will remain unconvinced that there is an actual problem of users producing incorrect pronunciations from our transcriptions as a result of reading our diaphonemic system as phonemic.

We are always open to making the key more decipherable, so if you have some advice on how to improve it or areas where it is confusing, please elaborate. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 21:32, 14 April 2015 (UTC)

Sorry, but this just makes it worse. There was not "one issue", there are many. The BATH and CLOTH issues were handled better before, when two pronunciations were given. Giving one, non-standard, representation that is supposed to represent two (but gives no indication as such), is absurd - no-one who doesn't think to click the link (and why should they, if they know IPA already?) is going to be anything other than hopelessly misled. W. P. Uzer (talk) 13:08, 16 April 2015 (UTC)

I think you're guilty here of dodging my point. Presumably, you are motivated to come here because you are concerned about a practical issue. Readers, according to you, will be "misled" into making the wrong pronunciation. In the previous discussion, I could barely get you to focus on providing actual concrete examples where this even could happen. You provided one (the BATH vowel) and we incorporated it into the system.

The more you dodge this point, which is the very crux of your argument, the more it will seem like your primary concern is theoretical, rather than practical. If that really is the case (and, at this point, I feel like our intelligence is being challenged by you pretending it isn't theoretical), then you're just wasting our time. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:22, 16 April 2015 (UTC)

You continue to miss my point completely. I have always given several examples of where there is likely to be a problem (see my numbered list below; I could add to it further). "Incorporating the BATH vowel into the system" makes things worse, not better, because it's the system that's the problem - BATH words were handled better when they were outside the flawed system. I've explained at great length WHY the system is flawed - can you either answer my arguments, or help work together to repair the problems that the system obviously has? W. P. Uzer (talk) 07:44, 17 April 2015 (UTC)

I haven't missed your point at all. I have already addressed the other examples you gave; for readers to actually mispronounce words as you have argued they would, speakers would have to go beyond their own accent, pronouncing vowels that don't exist in their repertoire or pronouncing words in ways that defy the phonotactic restraints of their dialect. That's simply not a realistic risk. You are arguing that, for example, an RP speaker will see a postvocalic r in our system and actually pronounce it. Does anyone really believe that will happen? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 08:50, 17 April 2015 (UTC)

Not everyone is a speaker with native-like competence in a specific dialect, nor should we assume that people are only interested in their own dialect. W. P. Uzer (talk) 06:16, 18 April 2015 (UTC)

Our target demographic is people who look at a word in our articles and wonder, "how does one pronounce that?" Our system is not designed to satisfy the amateur dialectologist's curiosity on local or regional pronunciations. In other words, yes we actually should assume that people are only interested in their own dialect.

This is the first time you've mentioned non-native readers. I can concede that the diaphonemic system would not be as ideal as what you propose for English learners below a certain level of competence. But the English Wikipedia's target demographic is people who are proficient in English. It doesn't make sense to me to scrap the current system just because non-native readers would have difficulty with it. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:23, 18 April 2015 (UTC)

I'm not proposing scrapping the system, just tweaking it. I don't think we should try to guess who our readers are and what information they want, and leave anyone who doesn't fit our paradigms out to grass. Anyway, the problems do not affect only non-natives - it's very hard for even a native speaker to understand what information is being conveyed sometimes. W. P. Uzer (talk) 07:40, 21 April 2015 (UTC)

As a person who can follow along on dictionary transcriptions (such as Merriam-Webster's or AHD's) but never took linguistics classes (which is one [large] category of "average" user), I must comment here that I find Wikipedia's transcriptions, made according to Help:IPA for English, to be perfectly cromulent, and just as good as the aforementioned brand-name ones. And the tooltip upon hovering makes it quite helpful. As a concrete example, when I view the article on bupivacaine and it tells me /bjuːˈpɪvəkeɪn/ , and it leads me by the nose as I hover ('b' in buy, long 'u' in 'cute', primary stress follows, and so on), it tells me everything I need to know, and completely satisfies me on the topic. W. P. Uzer seems to be someone with graduate/postgraduate linguistics expertise. What I would reply to them is not "you are wrong about the technical/linguistics aspect of your argument" (because I'm unqualified to judge that) but, just as importantly, this: as one among the large numbers of people who are supposedly being perplexed and misled ("largely undecipherable", "terribly wrong and misleading"), I honestly can't tell what it is on this page that I'm supposed to find incomprehensible, or even difficult. For example, the page tells me, "If the words illustrating two symbols sound the same to you (say, if you pronounce cot and caught the same, or do and dew, or marry and merry), you can ignore the difference between those symbols" ... "Footnotes explain some of these mergers" [if someone wants more information] ... To sum up what I'm trying to say, maybe it's true that some people who deeply understand phonetics and the IPA won't like how the IPA is being used here ... but that doesn't translate to large numbers of people being confused or unhelped by this page. Probably less than 1% of the general population can even understand the argument and counterargument about IPA use choices or schools of thought. The rest of us just want to know "'b' in buy, long 'u' in 'cute', primary stress follows," and so on. And people who use dictionaries with some frequency can get used to the conventions thereof and read the transcriptions directly; I have grown used to the system given here, as well as AHD's, for example, so as to sight-read them. Quercus solaris (talk) 22:24, 14 April 2015 (UTC)

But you are one of the lucky ones who has found this page. Most people who know IPA won't bother to click here. I'm not denying that a large majority of the transcriptions are perfectly OK, but some of them (particularly those with rhotic-only "r"s and variable vowels) will only be comprehensible (at least in those aspects) if someone happens to realize that what we are doing is not what we look as if we are doing (i.e. using IPA in the normal way) and clicks through and spends the time and effort to read this page. W. P. Uzer (talk) 13:08, 16 April 2015 (UTC)

You have absolutely no basis to assume any of what you say here is accurate. I think it's also clear from Quercus's comments that you have no authority to speak for people whose perspective you have no real understanding of. It's arrogant and rude. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:22, 16 April 2015 (UTC)

Huh? I'm pointing out that Wikipedia is putting inaccurate information in its articles. Ordinarily we correct such matters without having to demonstrate that any actual person has read the information and been misled by it. The fact that some people have not been misled by it is no excuse for retaining the erroneous information. I don't see how I'm being arrogant or rude to anyone, apologies to anyone if I inadvertently have been (though I make no apologies for being rude about the manifestly bad system that seems to be in use). W. P. Uzer (talk) 07:44, 17 April 2015 (UTC)

Actually, that's not a fact. It's an assumption on your part and a pretty specious one, considering that you have not even demonstrated how being misled by the system is even plausible. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 08:50, 17 April 2015 (UTC)

I have demonstrated that at length, see below, above and elsewhere. W. P. Uzer (talk) 06:16, 18 April 2015 (UTC)

Sorry, but you know that I've rebutted all your attempts to demonstrate this. That's the point. You've made your case and you've failed to convince. Thus, no demonstration. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:23, 18 April 2015 (UTC)

Your "rebutting" just seems to consist in narrowing down the target audience and their assumed mindset so as to try to exclude anyone who I demonstrate might be misled. Everyone seems in broad agreement (and this was the case in our previous discussion as well) that the potential to mislead that I've pointed out is a genuine one. W. P. Uzer (talk) 07:51, 21 April 2015 (UTC)

I am, of course, also referring to the previous discussion where you made more of a case for your position. There hasn't been much rebutting in this conversation (though, if you had been paying attention, you would have seen more than you mention; I'm assuming that you are once again being unnecesarily hyperbolic) because you haven't been presenting any evidence. The only actual argument that you've presented beyond "gah! it's misleading" is that the presentation of post-vocalic /r/ implies that they are pronounced in non-rhotic dialects. Since I'd already rebutted this in the previous discussion, I didn't think it was worth responding to. Repeating your argument doesn't make it stronger.

Beyond this, it's just been baseless repetition. I hope you don't consider your comments about Sudan, Boston, and diphthong as counting as arguments. You just say that they're misleading and nonsensical but don't explain why. Again, simply repeating your claim doesn't suddenly turn it into evidence, which is what would be required with a demonstration. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:49, 22 April 2015 (UTC)

I don't think my claim is in any way obscure or controversial. I'm pointing out that Wikipedia is using "IPA" symbols in ways which are not guessable either from the definition of the IPA or from the standard ways in which IPA is used by everyone else. Therefore readers won't know what we mean (and indeed, will think we mean something else, since the symbols we use do have meanings under that definition and those standard ways). I don't think that requires any more demonstration - it should be totally obvious to all of us. W. P. Uzer (talk) 07:52, 23 April 2015 (UTC)

Your claim, that people will be misled, depends on something that is quite controversial: that people untrained in linguistics or dialectology will produce features not present in their own dialect simply because they see a pronunciation indicated with those features. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:06, 23 April 2015 (UTC)

That's not what I'm claiming, but: people might produce the wrong features for their own dialect (if we indicate as long a vowel that they have as short, for example); people might not have an "own dialect"; etc. etc. as I've repeated already. W. P. Uzer (talk) 05:53, 24 April 2015 (UTC)

I'm interested in hearing of potential examples where our indication would prompt someone to produce a long vowel when it's short, but Boston, Sudan, and diphthong don't do that. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:03, 24 April 2015 (UTC)

Well, they might in all three of those cases, since some dialects have short and some have long vowels in those positions, and we mark either a short vowel only or a long vowel only. So some people, if they are foolish enough to trust us, are going to get it wrong. W. P. Uzer (talk) 07:58, 26 April 2015 (UTC)

How, exactly? What are the possible vowel confusions that a speaker of, say, RP would make? Or a speaker from California? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:14, 26 April 2015 (UTC)

For example, someone will see Sudan marked with a long a, and assume that means their long a, when their dialect actually has an ash there. W. P. Uzer (talk) 17:22, 26 April 2015 (UTC)

OK, so a speaker from California might read /aː/ as /ɑː/. That makes some sort of sense. That's one of the main reasons why people have been so reluctant to introduce this. The ideal symbol would probably be /a/, but I think people didn't prefer that one for some reason. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 18:11, 26 April 2015 (UTC)

The "ideal" would be not to use a symbol for this at all, and instead write out the different transcriptions clearly using IPA in the standard way. If we are to use a symbol, it must be something like my suggested Â, that doesn't look like anything in IPA. As soon as you start using standard symbols in non-standard ways, you're going to mislead and confuse some people. (/a/ would be liable to be read as ash by RP speakers, for example.) W. P. Uzer (talk) 08:59, 28 April 2015 (UTC)

I suppose, then, that there's no ideal solution for this one. As you indicated in the previous discussion, when there are two pronunciations, it's not clear who pronounces which one. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:28, 28 April 2015 (UTC)

W.P., there is no standard IPA transcription for English, as you suppose. Different British dictionaries use different IPA conventions, just as different Usonian dictionaries use different non-IPA conventions. — kwami (talk) 23:34, 14 April 2015 (UTC)

There is a pretty well fixed IPA transcription used in the major modern British dictionaries (pace a few minor details). If Wikipedia is going to use IPA, it should stick with that, because most people who know IPA will be expecting us to. In fact, it deviates quite markedly, but not in such a way that most users with a vague acquaintance with IPA (either from those dictionaries or from other sources) would realize that it is doing so. W. P. Uzer (talk) 13:08, 16 April 2015 (UTC)

Sure, for non-rhotic accents. But this is not Non-Rhotic Wikipedia. We don't differ much from British standards at all, apart from adding /r/ where it appears in rhotic accents. — kwami (talk) 18:56, 17 April 2015 (UTC)

And thus turning us into Rhotic Wikipedia, which is almost as bad. A simple measure like putting the r's in parentheses would repair this problem. Otherwise, the fact that we don't differ from British standards is in some cases a problem, and in some cases the fact that we deviate from British standards in a non-standard-IPA way is the problem, as outlined in more detail below. W. P. Uzer (talk) 06:16, 18 April 2015 (UTC)

@W. P. Uzer: It's true that the Wikipedia diaphonemic system isn't accurate for any one variety of English, and it would be very misleading if someone tried to learn how to pronounce English from it, since the actual pronunciations in a given variety are often different from the pronunciation you'd get if you literally pronounced the IPA symbols. However, it's good at representing most of the possible phonemic distinctions that occur across dialects, so that, as Quercus solaris says, a native speaker can use an {{IPAc-en}} transcription to figure out how they would say a word.

I personally think the diaphonemic system should be augmented by dialect-specific phonetic transcriptions, which can be used in articles that have an association with a particular dialect of English (say, articles on cities). For instance, it would be nice if Glasgow had a phonetic transcription of how a Scottish English speaker would say the word. Currently the only way to figure out how a Glaswegian would say the word is to look at the International Phonetic Alphabet chart for English dialects and see what IPA symbol is in the /oʊ/ and /ɑː/ rows and the ScE column. For the convenience of readers, it would be nice if the intro went like this —

Glasgow (/ˈɡlɑːzɡoʊ, ˈɡlæz-/ GLAHZ-goh or GLAZ-goh; Scottish English [glɐ̟zgoː]...)

or if the ScE pronunciation were provided in a footnote off the intro. However, this transcription might only be useful for readers who have specialized IPA knowledge (like me) — unless we had a Help:IPA for Scottish English page, or if the Scottish English page had audio examples of vowels. And not sure how much interest readers have in the actual local pronunciation of things, or if they're just interested in how to say things in their own variety of English. The diaphonemic system certainly helps with the latter. — Eru·tuon 04:29, 15 April 2015 (UTC)

Different readers are going to be interested in different things. First of all, though, we should stop telling blatant lies to people, by making non-standard use of IPA without telling them that we are doing so. Diaphonemes would not be a problem, as long as they aren't made to look like something else. I certainly think that for place names, local pronunciations should be given (possibly in footnotes) if we can source them. But I find the following things absurd as they are now:

Rhotic pronunciations given for place-names in England, implying that the "r"'s are pronounced (in either the local dialect or in standard RP, depending on how the reader interprets it). What's so hard about putting the "r"'s in parentheses, so at least the reader will get the idea that we're not making any statement about their presence or absence?
Similarly, British pronunciations for place-names in the US (and so on). Like Boston. We should prefer the relevant standard (GAm in this case), with other pronunciations (RP, local dialect, if sourced) given additionally or in footnotes. Trying to compact the information just leads to absurdly misleading results, with no particular gain except a tiny bit of space.
Sudan. This was way better before, when it was clearly indicated that there are two possible pronunciations, and that one of them corresponded to the sound file. (More information would have been nice about which country's dictionaries give which pronunciation, of course.) Now we have something that's utterly nonsensical to anyone who has not perused this help page in its fine detail, and still not particularly helpful to anyone who has (is the distribution of vowels in "Sudan" asserted to follow the same isoglosses as "bath", or is it just that there are the same two different vowels?).
Diphthong, similarly. Less disastrously, since the second vowel is not the motivation for giving the pronunciation, but still it's wrong to mislead people with some IPA knowledge about the way that vowel is pronounced.

To sum up, I would put rhotic-only "r"'s in parentheses, and be prepared to give more than one pronunciation in other cases where there is significant variation between the major varieties of English (at least RP and GAm, and the national standard and local dialect(s) in case of place-names and other terms with local associations). W. P. Uzer (talk) 13:27, 16 April 2015 (UTC)

And I will just point out yet another damaging aspect of the current system - people who have limited knowledge of IPA but who DO know the pronunciations are likely to mislearn IPA, and misuse it elsewhere, as a result of our non-standard use of it. W. P. Uzer (talk) 07:51, 17 April 2015 (UTC)

I’ve been trying to keep out of this argument, as there simply isn’t a correct solution to the problem of representing a variety of English accents with a single transcription. However, I’d just make a few observations.

Although the idea of the diaphoneme is not new, the idea of using a diaphonemic transcription for the whole English lexicon is, as far as I know, novel.
Since the diaphonemic transcription is not a standard phonemic transcription, I don’t think it’s a good idea to use slant brackets. Using a different type of bracketing would alert readers to the fact that the transcription needs a special type of interpretation.
If I’m right that the WP diaphonemic transcription is novel, it worries me that this could be seen as a case of OR.
In my ideal world, each transcription would link to an audio playback to allow readers to listen to an authentic pronunciation. The obvious example of this is the Forvo site: in the case of ‘Glasgow’(quoted above), you can find on Forvo about a dozen British pronunciations of the name, and 3 Scots pronunciations. Unfortunately, a lot of Forvo recordings are either of poor technical quality or spoken by people who seem unused to using a microphone. RoachPeter (talk) 15:07, 17 April 2015 (UTC)

@RoachPeter: Different bracketing would certainly be good. Diaphoneme § Representation lists several options: //double brackets//, !exclamation points!, |vertical bars|, and {curly brackets}. I don't know which of these is used more frequently. Vertical bars would be troublesome, because they are a part of template syntax and would require using the template {{!}}, and curly brackets would be troublesome because double curly brackets are used in template syntax. However, any of these could be coded into the template {{IPAc-en}}, and thus automatically included in transcriptions. — Eru·tuon 17:33, 17 April 2015 (UTC)

A change of bracketing would be a small improvement, but would hardly address the problem - close to 100% of readers would not notice or understand the difference. W. P. Uzer (talk) 07:06, 18 April 2015 (UTC)

Unless you've taken a poll of readers' reactions, there's no basis to claim that 90 or 99% of them would do this or that. People tend to vary in how attuned they are to subtle formatting choices. Some people notice these things, some don't. And some people will never understand the difference between phonemic and phonetic representation either, or notice the bracketing differences (square brackets and slashes).

If diaphonemic representations are too phonetically misleading, phonemic and phonetic representations may be as well, and if bracketing around diaphonemes is too hard for readers to notice, then bracketing around phonemes and phones may be as well. Unless diaphonemic representation is more misleading to readers than phonemic or phonetic, and diaphonemic bracketing is less noticeable to readers than phonemic or phonetic, your argument may actually apply to all three: diaphonemic, phonemic, and phonetic bracketing and transcriptions. — Eru·tuon 22:26, 18 April 2015 (UTC)

Yes, it may, my point is basically the same as it always has been - we should do what people are used to, what is effectively a standard (insofar as there is a standard) in the real world. Most of Wikipedia's transcription system meets this requirement, but there are a few aspects where, as I've pointed out, it fails, and that to me is clearly wrong. W. P. Uzer (talk) 07:47, 21 April 2015 (UTC)

I was going off-topic mentioning transcriptions or phonetic symbols. My main point was regarding bracketing. You said most readers won't notice different brackets, but you haven't given any evidence for that assertion. And even if they don't notice it, that's not a reason not to use it, since we already use different bracketing for phonemic and phonetic transcriptions, and some readers will not notice or understand the difference between slashes and square brackets. Since we have phonemic and phonetic transcriptions marked by two types of brackets, it's legitimate to have diaphonemic transcriptions marked by a third type of bracket. — Eru·tuon 21:19, 21 April 2015 (UTC)

My "evidence" would be the general experience of the linguistic knowledge of the people we encounter in everyday life. They have little clue about phonemes, phonetic transcriptions, diaphonemes, and still less about the different types of brackets that are conventionally placed around them. I think you would have to spend all your life in a university linguistics department not to realize this. W. P. Uzer (talk) 07:52, 23 April 2015 (UTC)

I haven't spent all my life, or even a significant part of it, in a linguistics department, and I can't argue that you're wrong about ordinary people's knowledge of linguistics. I was talking about readers' ability to pick up on detail, which is unrelated to level of linguistic knowledge. But we're both giving our personal impressions, and I'm not sure if I disagree or not, so let's leave this question behind. What about my other point? As I understand it, you're saying that we shouldn't use something readers won't notice or understand, so we shouldn't use separate bracketing for diaphonemes. But, as you say, readers won't notice or understand phonetic and phonemic bracketing, so your argument applies equally for all three: phonetic, phonemic, and diaphonemic bracketing. Why is diaphonemic the only bracketing we shouldn't use? Or am I misunderstanding your argument? — Eru·tuon 09:07, 23 April 2015 (UTC)

I'm not objecting to using diaphonemic bracketing - we should use it if we are using a diaphonemic system. But I'm objecting to using a diaphonemic system - or at least, a diaphonemic system that looks almost the same as the phonemic or phonetic systems that people are used to. I don't think the use of different brackets is sufficient to alleviate the confusion - it's not a detail that real people would pick up on, particularly as they almost certainly don't know the conventions that professional linguists use. W. P. Uzer (talk) 10:05, 23 April 2015 (UTC)

@RoachPeter: You're wrong about diaphonemic transcriptions being novel. Merriam Webster dictionaries have done that for decades. I also recall them using back slashes. I don't know if that's a meaningful distinction from virgules or simply an error. And, as I noted above, The Cambridge Grammar of the English Language uses diaphonemic transcriptions. — kwami (talk) 18:40, 17 April 2015 (UTC)

Also, our policy on original research is much less stringently applied to in-house conventions. In other words, we apply WP:OR to the pronunciations we provide, not the way we indicate them. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 05:52, 18 April 2015 (UTC)

I think it's understood as a given that we indicate facts in ways that are broadly established in the real world. Otherwise we are not "indicating" them at all. W. P. Uzer (talk) 06:30, 18 April 2015 (UTC)

We're talking about WP:OR here, W.P. My point is that the rule doesn't necessarily apply to in-house conventions. Whether a convention is confusing or flawed is tangential to my point. Talk page discussions have a way of splitting and winding down tangential avenues from the O.P.'s original intent. You don't need to pop in and try to redirect them back. You're coming off as immature or controlling. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:23, 18 April 2015 (UTC)

Please don't make personal attacks, we're all presumably on the same side here, just tossing ideas around to try to make things better. If others can be tangential, then so can I be. W. P. Uzer (talk) 07:54, 21 April 2015 (UTC)

Don't consider it an attack. You are indeed in your right to say what you wish. I was letting you know the impression you were giving off. If you don't care, that's your problem I suppose. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:49, 22 April 2015 (UTC)

@W. P. Uzer: I agree that we should stop blatantly lying to readers, or at least misleading them that this diaphonemic system uses the literal meanings of the IPA symbols. A note should be added at the top of the page to clarify what the system actually is. Edit: I see you already added a note. Excellent.

It's fair to note that the system can cause people to mislearn the IPA. However, other systems can do the same: for instance, a British IPA system that uses [æ] would mislead speakers in England who actually have the open vowel [a], and there are other cases like this. No IPA system would be perfect, partly because of changing pronunciations and because of different accents. — Eru·tuon 17:41, 17 April 2015 (UTC)

That's true, but we should at least stick to the IPA-based system that is used in the real world, and not unnecessarily and misleadingly introduce new symbols that look like IPA symbols representing a particular sound, but are actually used by us to represent two specific different sounds. How can anyone possibly think it's better to do that? I know we want to present the information in a reasonably compact form, but not to the extent that we cease presenting the information at all, and instead present different, false, information. Surely it is obvious to all why this is wrong? W. P. Uzer (talk) 06:30, 18 April 2015 (UTC)

Yeah, the /aː/ and /ɒː/ symbols are particularly strange, but there are two ways in which they make sense. Phonetically, the symbols are basically intermediate between the sounds they represent in modern RP and GA, and they have historical or dialectal basis. In an earlier period of Modern English, the vowels of trap and bath, and lot and cloth were [a] and [ɒ], but then bath and cloth were lengthened to [aː ɒː] in certain dialects, and later shifted to [ɑː ɔː] (giving the values in RP and GA respectively), while trap and lot were fronted and left unchanged, yielding [æ ɒ]. Also, bath has [aː] in some English dialects, I think those that are intermediate between Northern and Southern England English. These phonetic and historical justifications aren't given on the page, and I'm just improvising here, but I think these may be the reasons why the symbols were chosen. — Eru·tuon 07:41, 18 April 2015 (UTC)

Yes, I understand all that, but it is still wrong in practice to use them, since they fail to convey the information that we intend them to convey. It would be less bad if we used a symbol that looked like nothing on earth, like / *Â* /, then at least people would be alerted to the fact that something strange is going on. But still it would be a solution to an imaginary problem - there's no reason why we should need or want to give a single representation in all cases. W. P. Uzer (talk) 10:34, 18 April 2015 (UTC)

It's true there's no reason we have to, but it is neater to have a single symbol rather than two different ones. It allows there to be a single transcription in articles rather than two. — Eru·tuon 22:26, 18 April 2015 (UTC)

Yes, it's neater, but that isn't any kind of excuse for presenting what is effectively untrue information. The top of this page trumpets (rather exaggeratedly, in my opinion, but that's another story) how IPA is superior to other transcription systems because people around the world generally know it; then we turn round and use IPA in a way that differs markedly from what these people around the world presumably know. It's just absurdly illogical in my opinion. W. P. Uzer (talk) 07:35, 21 April 2015 (UTC)

I don't see where the top of the page trumpets the superiority of the IPA. Yes, the symbols /aː ɒː/ are inaccurate and misleading, if taken literally as representing a long front or central open vowel and long rounded back open vowel. But if phonetic accuracy is a goal, then there are many other problems with the diaphonemic system, and with the systems used in dictionaries. For instance, the traditional RP system is inaccurate to varying degrees for modern speakers in South East England (or so I from Geoff Lindsey). The symbols /eɪ e iː æ ɒ ʌ ɔː ʊ uː aɪ aʊ/ all have problems, because of sound changes and rising prestige of Estuary English.

The question is whether another IPA system would be more phonetically accurate, how much more accurate it would be, how many English speakers it would be accurate for, which English speakers it should be accurate for, what dialectical phonological differences it represents, which dialectal differences it glosses over, and how much it corresponds to the IPA systems used for other of the most-spoken languages. Personally, when I look at the International Phonetic Alphabet chart for English dialects, I'm not sure if phonetic accuracy is possible or if it should even be a goal. — Eru·tuon 21:19, 21 April 2015 (UTC)

What exactly do you mean by "misleading"? In the context of our pronunciation guide, even ⟨aː⟩ and ⟨ɒː⟩ wouldn't be misleading as long as they don't prompt readers to produce an incorrect pronunciation. I'm pretty sure it isn't misleading in that sense.

Taking W.P.'s example of Sudan being transcribed as /suˈdaːn/, a reader from California will read it as [suˈdæn] while a reader from Australia would read it as [sʉːˈdæːn] (or [sʊˈdæːn]. I don't know how Australians pronounce bedroom).

Even a reader proficient in the IPA but not in our diaphonemic system (a problem W.P. believes to be much more frequent than can be proven) and who attempts to read our system phonetically wouldn't be misled, because people don't suddenly produce vowels that are not in their system. If an IPA-literate reader from California saw the transcription /ˈbɒstən/, they wouldn't go about pronouncing the word with [ɒ], because that's not part of their vowel inventory. They would do what people from California normally do when they hear words pronounced with [ɒ]: replace it with [ɑ]. Similarly, an IPA-literate reader from London would read /ˈɡlɒstər/ the same as if they had heard someone say [ˈɡlɒstəɹ]: they might take note of the /r/, but they will pronounce it [ˈɡlɒstə] because they speak with a non-rhotic accent where post-vocalic /r/ is not present. They would not begin pronouncing pronouncing post-vocalic /r/ simply because we put it there in our system.

I suppose it might be misleading in the sense that people might get a skewed idea of what the vowels actually represent in IPA. The reader from California might believe that ⟨ɒ⟩ represents an unrounded vowel, for example. But, as I hope it's been demonstrated already in Erutron's recent comment above, IPA-literate English speakers already get misled in this way. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:34, 22 April 2015 (UTC)

Yes, it's certainly unfortunate that the standard way that dictionaries use IPA deviates quite a bit from the official IPA scheme, but there's little we can do about that. But that's no excuse for making things even worse, by deviating from both of those systems, while still making the symbols identical to something in official IPA (and similar or identical to something in the dictionary systems). I think it's prima facie obvious how this is misleading and wrong. It's not necessary to provide "evidence" that when people see something that means X not Y in all known and standard ways of conveying information, they will think it means X not Y when they see it in Wikipedia. No evidence is provided either for the claims about how people from California (etc.) will react - it's rather absurd to imagine they will always react in any particular way. As I've already said, not everyone is a native speaker of any dialect (or of any of the dialects the current system attempts to cater for), not everyone's interest is restricted to their own dialect, and certainly not everyone will guess that Wikipedia's scheme is supposed to represent multiple dialects or guess the system by which it attempts to do so. Meanwhile the changes I've suggested are fairly minor and don't seem to have any serious downside. My aim is not full phonetic accuracy or any other unachievable goal, just to align Wikipedia with the real world in such a way that we can feel that the information we are providing is actually true. W. P. Uzer (talk) 07:52, 23 April 2015 (UTC)

To summarize, the problem is that /aː ɒː/ don't represent their literal phonetic values, long open central and long open rounded back vowels, but are symbols representing two different phonemes in different dialects. And if a reader interprets the symbols as literally phonetic, they'll be misled on what the IPA symbols represent, and be confused when they see the same symbols used in other languages. That's a problem, but there are two ways to solve it: either don't use the symbols, or tell readers that they aren't literally phonetic. At the moment, the page does the second (though it could be clearer and more direct), but you want to switch to the first. A sufficient reason for switching would be that readers are confused, but to this point nobody's actually been confused; you've only said they are likely to be confused. There's no point arguing over whether readers will or will not be confused; it's basically an attempt at mind-reading. Perhaps the best way to settle this would be an RfC — ask editors and readers whether they are confused by /aː ɒː/ or not. — Eru·tuon 09:07, 23 April 2015 (UTC)

I don't think it's even necessary to do that; we can see that anyone who doesn't think to click the link through to this page is going to be confused - there's no way they could possibly not be. It's just not acceptable for Wikipedia to say something that isn't true, on the grounds that people will click a link that tells them that we are presenting information in an idiosyncratic way. It's exactly like saying "Barack Obama is a [Help:Political designations|socialist]", where the Help page explains that on Wikipedia, "socialist" is taken to include US Democrats. Such nonsense would be immediately expunged without anyone having to show that anyone has actually been misled by it. W. P. Uzer (talk) 09:58, 23 April 2015 (UTC)

WP, do you believe that an IPA-literate speaker from California (who doesn't have /ɔː/ or /ɒ/ in her vowel inventory) will/could/would actually pronounce Boston with [ɒ] after seeing our transcription at Boston? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:06, 23 April 2015 (UTC)

No, but we've no idea how they actually would pronounce it, and that kind of situation is not my only concern anyway, as I've already said several times. W. P. Uzer (talk) 06:15, 24 April 2015 (UTC)

Do you believe that a speaker of a non-rhotic dialect, such as RP, would pronounce a post-vocalic /r/ after seeing our transcripion at Glaucester? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:03, 24 April 2015 (UTC)

We've already been over this - the answer is the same as my immediately preceding one. W. P. Uzer (talk) 07:58, 26 April 2015 (UTC)

So if such a person were to read our transcriptions and produce an incorrect pronunciation because there is a post-vocalic /r/ indicated, what are some possible mispronunciations that you envision? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:14, 26 April 2015 (UTC)

I don't know, possibly in this situation it isn't so likely that a native will mispronounce anything, but a non-native might, and everyone will potentially be misled about what Wikipedia is trying to tell them (I would think that it's telling me that the local dialect around Gloucester is rhotic). W. P. Uzer (talk) 17:22, 26 April 2015 (UTC)

Our goal is to indicate to native speakers how to pronounce a word. If transcribing postvocalic /r/ doesn't steer them away from the intended pronunciation, then it's not misleading. Mispronunciations are the only kind of misleading we should be concerned about, not dialectal information. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 18:11, 26 April 2015 (UTC)

I absolutely do not accept that - Wikipedia should be providing accurate information for everyone, not just a selected subset of readers who we assume to be making use of the information in a certain limited way. (The reason given at the top of this page for using IPA is that "for most people around the world, the IPA is far less confusing" - so apparently we are aiming to serve people around the world. If we just wanted to serve native speakers, then probably respelling would be a way better option in a lot of cases.) The current system doesn't even serve your goal particularly well anyway, as has been pointed out elsewhere in this discussion. W. P. Uzer (talk) 09:14, 28 April 2015 (UTC)

Your argument is getting more and more muddled. What, exactly are you not accepting? That our goal is to indicate pronunciation? That we should not be concerning readers with dialectology? That transcribing postvocalic /r/ doesn't lead to an incorrect pronunciation? That our primary target is native speakers? And when you say that Wikipedia should be providing accurate information, how is the system providing inaccurate information beyond the people who misread it? What, beyond pronunciation, should we be giving in our pronunciation guide? How is it that not focusing on whatever it is you think we should also be focusing on is "limited"?

By the way, a little bit of linguistic geography for you: native English speakers live all around the world. In addition to the US and UK, there's also South Africa, Australia, New Zealand, and India. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:28, 28 April 2015 (UTC)

I think I've made it abundantly clear what I'm not accepting - the use of standard symbols to mean something different from what they standardly mean. People who interpret such symbols in the standard way are not misreading them - it is we who are misusing them. So it very much is not acceptable that such people should be misled. (Or that other people should be mistaught about what the standard symbols mean.) W. P. Uzer (talk) 19:52, 28 April 2015 (UTC)

This is what I mean by muddled. You said, "I absolutely do not accept that." In common parlance, such a statement is interpreted to refer to something immediately stated. While it is true that your general stance has been clear, nowhere in my immediately previous post did I state that we should use IPA symbols in such and such particular way. Was there actually something that I said in the post immediately previous to when you said "I absolutely do not accept that" that you disagree with or are you losing track of the thread and restating your general position just to restate it? I'm trying to have a conversation here, and you seem to keep getting lost. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 00:55, 29 April 2015 (UTC)

I was disagreeing with what seemed to be your overall assertion: that the misleading of certain readers, or of all readers in certain ways, doesn't matter, because we've defined a very specific goal and don't care about anything or anyone else. (I also happen to disagree that the system currently in use is even good for achieving that specific goal.) W. P. Uzer (talk) 22:46, 29 April 2015 (UTC)

As with the symbols /aː ɒː/, we need to see if readers are actually confused by the /r/ in Gloucester. Readers might or might not be confused, and unless we know that they are, we don't have a clear reason to put /r/ in brackets. — Eru·tuon 01:07, 30 April 2015 (UTC)

As I've said before, we don't know and really can't know one way or the other what the vast body of users are doing with our information - we just have to strive to make the information as accurate as possible, based on what reliable sources tell us. We certainly can't excuse saying something that is not true on the grounds that we don't have solid evidence that anyone believes it. Stating that "the pronunciation" contains an "r" sound" is less accurate than saying that the pronunciation contains an "r" sound for some people but not for others (which is still not the whole story, but at least is not a lie). W. P. Uzer (talk) 03:23, 30 April 2015 (UTC)

(edit conflict) OK, so what WP disagrees with is that our primary goal for a pronunciation guide is to indicate pronunciation. Irrespective of the merits of WP's ideas in addressing the issues they are concerned with, I think we'd need to get other people on board with actually prioritizing something other than indicating pronunciation in our pronunciation guide first. In other words, you may be concerned about dialectological issues and non-native speakers below a certain level of proficiency, but if those concerns come at the expense of clearly indicating pronunciation, then they aren't worth considering. I understand that you also believe that clearly indicating pronunciation is not very well done, but so far getting you to elaborate on that has been like pulling teeth. I think it would be in your interest to focus on the areas where you believe this system fails in its primary goal, rather than assuming (wrongly) that your peripheral goals are shared by everyone else. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 03:31, 30 April 2015 (UTC)

Of course the system should be indicating pronunciation, I don't know where you got the idea that I suggested it should be doing anything else. But it should be doing it in an objective and accurate way. I've been elaborating on it over and over, explaining very clearly why the present system (in a few respects) is not accurate, and at least on some points we seem to have reached agreement about that. I certainly am concerned with non-native speakers as well as natives, as we all should be; but I've also shown that natives aren't served well by the present system either (giving one non-existent pronunciation when we mean two different ones is a simply bizarre way of trying to convey information). W. P. Uzer (talk) 06:59, 2 May 2015 (UTC)

Are you seriously going to read what I said as meaning that I think that you you don't think we should be indicating pronunciation? I said primary goal. I'll repeat what I said: when other concerns (in your case, giving dialectological information and catering to non-native speakers below a certain level of proficiency) come at the expense of clearly indicating pronunciation (that is, getting our readers to pronounce words as we intend them to), then they aren't worth considering. That's how we measure accuracy; the concerns about objectivity are more theoretical (and ephemeral) than practical. We're being practical here. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:11, 2 May 2015 (UTC)

I really don't know what you mean now. We want to indicate pronunciation; no other goal has been suggested. But as I've demonstrated, the current system (in the cases under consideration) doesn't indicate pronunciation correctly. We can't just say that "foreigners don't matter", and we can't ignore the accepted meanings of IPA symbols and make up our own. This is neither practical or objective or accurate or anything, it's just wrong on all levels. W. P. Uzer (talk) 10:54, 4 May 2015 (UTC)

Nor can we avoid giving "dialectological" information when English has no single standard dialect, And nor should we want to avoid it - if I can find out from Wikipedia how a French person says "Paris", why should I be prevented from finding out how an American says "Boston", how an English person says "Gloucester", or how a Scot says "Glasgow"? W. P. Uzer (talk) 11:00, 4 May 2015 (UTC)

OK, which is it? You say you haven't suggested any other goals than indicating pronunciation, and then in the very next paragraph reiterate what you've been saying repeatedly: that we should be concerned about conveying dialectological information and catering to non-native speakers in our English pronunciation guide. Those would be additional goals on top of indicating pronunciation. You are incorrect when you say that we can't avoid giving this information, the system that we have does just that. It seems that you believe we accidentally indicate dialectological information, but your assumption that a significant portion of users would attempt to glean dialectological information from this system is pretty unreasonable. I don't think people would do that.

If there is a rule that says a local pronunciation can't be included in an article in addition to a diaphonemic one, then that's news to me. AFAIK, there's nothing stopping people from doing as such, apart from proscribing the parsing of national standards (we wouldn't give the diaphonemic pronunciation of Australia and an Australian one (not in the lede anyway).

You can close your ears and stick your tongue out all you want, but this is indeed a practical approach. As long as people understand the system, they will get to the right pronunciation and that's all we're concerning ourselves with. I understand the theoretical issue you have with using the IPA for a diaphonemic approach, but we have long shrugged our shoulders and accepted that as a necessary compromise. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 18:46, 4 May 2015 (UTC)

You seem to have a much more limited interpretation of the phrase "indicating pronunciation" than I think other people would (and certainly I do). To you it seems to exclude indicating any pronunciation to anyone except in the one special case where (a) the person is a native speaker (b) they want to know only the pronunciation in their own dialect (c) they are savvy or lucky enough to guess that they have to click the blue text in order to work out what some of the symbols mean (even though these are symbols that have a different meaning in the system which, according to the message at the top of this page as you interpret it, people around the English-speaking world are claimed already to be generally familiar with). And it doesn't matter to you that anyone outside this select group gets either no indication of pronunciation, or worse - indication of the wrong pronunciation. If people have shrugged their shoulders and accepted this rather ridiculous situation in the past (as a compromise? between what???), it's time to stop shrugging and start thinking about how best to put it right. W. P. Uzer (talk) 19:16, 6 May 2015 (UTC)

Yeah, we're focused on native speakers who are only interested in their own dialect. That's not an untoward focus. Most of our readers are native speakers (or sufficiently proficient in English to have learned a particular dialect, usually either RP or GA) and, when people look up pronunciations, they are generally concerned with pronunciation in their own dialect. There's nothing wrong with an interest in other dialects, but our pronunciation guide does not cater to that interest.

Plus, (a) and (b) are actually true of quite a few dictionaries. As for (c), I'm not sure what the remedy would be (if it is, indeed, a huge problem, which I'm not sure it is), but you haven't really focused on strategies to help readers to come to the guide or to make the guide clearer. Do you have any actual suggestions for that? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 19:35, 6 May 2015 (UTC)

I'll answer in a new section at the foot of the page, this is getting confusing. W. P. Uzer (talk) 19:37, 6 May 2015 (UTC)

@W. P. Uzer: Obviously we can't know how all readers and editors react to the diaphoneme system, but we can find out if some are misled, and how many respondents are misled. It's possible that something's inaccurate but nobody's misled by it. Misleadingness and inaccuracy are related, but not the same. Inaccuracy is a logical question of defining terms used and determining if they're used correctly, but misleadingness is related to people's reactions. The only way to determine people's reactions is by asking them. Arguing whether or not someone should be expected to be misled is pointless. So we need an RfC or something. Not sure what page to put it on. Æµ§œš¹, do you have any ideas? — Eru·tuon 01:36, 2 May 2015 (UTC)

This is absolutely not what we do on Wikipedia. I've never before seen it suggested that "we know this information is inaccurate, but before we change it, we have to find a certain number of readers who are actually misled by it." An RfC can never hope to reach all readers everywhere (WP editors are not a typical cross-section), and people who are misled are not normally aware that they are misled. W. P. Uzer (talk) 07:14, 2 May 2015 (UTC)

I didn't say you had to find readers who are misled by incorrect information. I said there are two separate questions: should we change the diaphoneme system because it's inaccurate? and should we change the diaphoneme system because it's misleading? The second question needs an RfC, while the first just needs some clarification on what inaccuracy means. — Eru·tuon 08:37, 2 May 2015 (UTC)

I'm not sure if an RfC would get at what we're trying to figure out. For one, the system presumably stops being confusing when people read the explanatory guide (WP has iterated that their concern is that people won't read the guide), so we'll have to find people who haven't read the guide, and then get them to reveal if they got the wrong pronunciation. With the /aː/-/ɑː/ issue, for example, any reader who misreads the former as the latter isn't necessarily going to even know that they have been misled. The most we can get is other people (many of them likely not versed at all in linguistics) speculating on what fictional other people might be confused at, which is what we've been doing all along. Not only would this be just as unconvincing as evidence, but it would bring in many people weighing in on a topic they're not knowledgeable in. We'd be inviting baseless speculation. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:21, 2 May 2015 (UTC)

I agree that such an RfC would not be particularly useful; no harm in asking people what they think, I suppose, but as Aeusoes implies, most people just wouldn't appreciate the complications of the topic, and people who have been misled might not know about it (and probably won't be reached by the RfC anyway). W. P. Uzer (talk) 10:54, 4 May 2015 (UTC)

WP: It seems you are presenting two arguments: first, that readers will be confused; second, that using the symbols is illogical. The argument from illogicality doesn't require evidence, but the argument from readers' confusion does. To this point, you've only said readers would or should be confused by the symbols, but haven't shown that readers actually are confused. Unless there's evidence that readers actually are confused, readers' confusion isn't a reason for not using the symbols. That's why I suggest an RfC. — Eru·tuon 02:24, 24 April 2015 (UTC)

But whenever we see something manifestly wrong in Wikipedia, we just correct it, we don't need to go through the highly difficult process of actually a finding a real person that was confused by the wrong information. (Was any such RfC conducted before the introduction of these non-standard multi-meaning symbols?) And surely you aren't disputing the assertion that if someone sees these symbols and doesn't think to click the link to this page (the only place in the universe where our invented meanings of these symbols is explained), they will not know what the symbols mean? And since the symbols do mean certain (other) things in the real world, is it not reasonable to assume that many readers will not click the particular link (since it can't possibly be readers' practice to click every bit of blue text they see, just to check that Wikipedia hasn't invented its own non-standard convention as to the meaning of a particular standard term or symbol)? W. P. Uzer (talk) 06:15, 24 April 2015 (UTC)

We wouldn't be looking for comment as to whether it was "wrong", but whether it was confusing or misleading. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:03, 24 April 2015 (UTC)

Forgive me if this reveals that there's something earlier in the thread that I have missed or forgotten about.
One question I have is this: If you took all the symbols in this system and replaced them, 1:1, with non-IPA symbols (leaving everything else alone), could you solve the aspect of this system that W. P. Uzer objects to?
If no, [what follows here?]
If yes, then the corollaries become as follows:
- What would you replace them with?
  - Would the replacements violate the principle of "Wikipedia should not use any system that is noticeably idiosyncratic/WP:OR"?
    - If yes, [what follows here? No viable path but "leave as-is"?]
    - Is there anything that could be done with macrons and breves (ā ē ī ō ū / ă ě ĭ ŏ ŭ) that would produce a "no" answer there (violate=no)?
      - If yes, should it be done? Would it be too "American"/US-centric? Would it generate a storm of annoyance?
I'm not sure whether this line of thought leads anywhere consequential. Worth sandboxing at least. Quercus solaris (talk) 22:44, 23 April 2015 (UTC)

Certainly it would be an improvement (but still not my preferred solution) to replace the non-standardly-used symbols with symbols that look nothing like anything in IPA (I suggested *Â* above). In fact that is sort of my preferred solution with the "r" - just put the "r" in parentheses. With the multimeaning vowels, however (other than those like /i/ which are already in common use), why on earth not state the alternative pronunciations explicitly? I can't see why anyone thinks it's kinder to readers to tell them that Sudan is pronounced "sʊˈdÂn" (whatever that is even supposed to mean - the explanation on this page for a: is not entirely clear anyway) when we could be telling them what we really mean, that it's pronounced "sʊˈdæn or sʊˈdɑːn" (with the possibility of adding more specific information from dictionaries to indicate which version is used where, etc. - in a footnote if it gets too long). W. P. Uzer (talk) 06:15, 24 April 2015 (UTC)

Does anyone actually disagree with this? To me it seems a blindingly obvious solution to something that should never have been a problem. W. P. Uzer (talk) 09:14, 28 April 2015 (UTC)

We already have a respelling system to be used alongside the IPA. Putting postvocalic /r/ in parenthesis would not solve any actual problems; it might actually be more confusing, as readers may believe that it's optional, which is not the case. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:28, 28 April 2015 (UTC)

Of course no system is perfect, but surely "optional" is closer to, or at least no further from, the truth than "compulsory"? Particularly in cases where the country to which the article relates is a predominantly non-rhotic one - in that case it seems we could practically not mention the "r" at all, or put a rhotic pronunciation in a footnote. W. P. Uzer (talk) 19:52, 28 April 2015 (UTC)

I'm not sure if "compulsory" is the right word, but for rhotic speakers it is indeed the correct pronunciation to have postvocalic /r/ where we indicate it, even for placenames in non-rhotic speaking areas and the names of non-rhotic speaking dialects. Parentheses would imply otherwise and relegating rhotic pronunciations to footnotes would not help either. As I stated above, non-rhotic speakers will not pronounce postvocalic /r/, even when we transcribe it without parentheses. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 00:55, 29 April 2015 (UTC)

I think we're just repeating ourselves now - I've already pointed out defects of the "always use rhotic" solution, but I'll list them again: non-natives (with imperfect academic knowledge of English phonotactics) will assume that the "r" is to be pronounced in all relevant dialects; anyone may assume (in a case like Gloucester) that the pronunciation being given is a local dialect one, since it is impossible in the national standard dialect; anyone may learn (wrongly) that standard use of IPA in English involves putting an "r" in in places where it is not pronounced. Of course just putting the "r" in parentheses doesn't make all possible problems and misinterpretations magically go away, but I think it's an improvement in most respects. If the result is still going to look too confusing, then we should perhaps adopt a different tack: always label our transcriptions as to which dialect they refer to (e.g. RP, GAm) - I'll flesh out that idea some other time than right now. W. P. Uzer (talk) 22:03, 29 April 2015 (UTC)

Speakers of non-rhotic dialects — I include myself here, as a speaker of a variety of General British — are perfectly capable of ignoring 'r' in non-prevocalic positions. We do it when guessing the pronunciation of words from their English spelling, and we can do it when interpreting pronunciations from an IPA transcription. Since /rC/ isn't allowed anywhere by our phonotactics, it is our natural instinct to just omit those /r/s. DavidPKendal (talk) 09:11, 1 May 2015 (UTC)

That's true, but as I've already noted, not everyone falls into that user category - non-natives may not know the phonotactic rules; even natives may want to find a pronunciation in a different dialect than their own; people may mislearn the standard(ish) way IPA is used by observing the way we do it; people who know IPA and the phonotactics of the main variety may conclude (in the case of e.g. place names) that the pronunciation being given is a local rather than a trans-Atlantic one (I think that was my own initial reaction). W. P. Uzer (talk) 07:14, 2 May 2015 (UTC)

Suggestions for improvements

Latest comment: 8 years ago23 comments9 people in discussion

I'm asked (see thread #Bad, above) to suggest improvements to the situation where readers who don't meet two particular conditions - i.e. that they are natives or native-like speakers and are interested only in their own dialect - are fed empty or false information (one editor seems to think that isn't a problem; I hope no-one else shares his view), and that even readers who do meet those conditions may be fed false information if they happen to believe, rather than click on, the transcriptions that they see. There seem to be several possible ways to go:

Stick to a diaphonemic system, but use different, non-IPA-like symbols, for the non-standard meanings. That way everyone will hopefully know that they have to click rather than apply their existing IPA knowledge. (Though there is still a danger that some people will think that these symbols are standard IPA, and thus mislearn the standard. And why make people click when we can just give them the information straightforwardly?)
Write out the different pronunciations (I suggest RP, GAm and any relevant local pronunciation would suffice) using standard British-dictionary-style IPA. It's a bit longer, but that's hardly an issue. Nor is it an issue that we would appear to be favoring these two "dialects" - we already do favor them (and we should, since they're the de facto international standards), we just hide the fact.
Switch to respelling rather than IPA. If our target audience really is native speakers, then this is a way more accessible solution (though it doesn't work for every word; we shouldn't use the Wikipedia respelling "system", either, if it really does tell us to respell "vice" as "vys" as the information at the top of this page seems to imply).
Stick to the current system, but change the phonemic slashes to more technically correct diaphonemic delimiters, and add an explicit link to "key" or something, so as to at least hint to people that they need to click (this is a bare minimum solution for me; I think still a lot of people will think "oh, this is just IPA, so no need for me to read the key").

My preferred solution is number 2, until someone can point out any significant downside. W. P. Uzer (talk) 20:02, 6 May 2015 (UTC)

A situation like #2 was actually the direct antecedent to the diaphonemic system. I think the major issue that users had was the clutter. Two pronunciations doesn't seem like much, but there's actually no good reason to limit to just RP and GA; even if you limit yourself to dictionaries, Australian English is a common additional pronunciations that users were including before we changed things. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 21:13, 6 May 2015 (UTC)

Regarding the length or clutter, I'll put in a plug here for gradually increasing the tendency that when the etymology and pronunciation info has any length to it, it is moved down to a section, with a link in parens at top. Two advantages: (1) a cleaner lede and (2) the etymology and pronunciation info can be as long as it needs to be—the length isn't getting in anyone's way. See examples at pedagogy and tretinoin. Mark Twain talked about the German who surfaced on the other side of the ocean with his verb in his mouth. That is how the opening of WP ledes sometimes feels. Before: "Foo (/fuː/ in Martian accent, /fʊ/ in Lilliputian accent, /fʌ/ in Atlantean accent; from the Greek φω, "placeholder name", akin to Latin fu, "whachamacallit"; also sometimes called foo substance, foo-ite, or Eurasian foo, the latter when differentiated from the Antarctic foo) is a type of bar found underground." After: Foo (etymology and pronunciation) is a type of bar found underground." Quercus solaris (talk) 01:34, 7 May 2015 (UTC)

Not only do I agree with you in principle, but I'd be even more aggressive. I'm guessing that 95% of readers of a given article don't go there for pronunciation or etymology, so expending 26 characters on that metadata feels like a lot; in the case of pedagogy, that link represents a tenth of the characters in the lead. I'd propose something like "Foo^[pron.] is a type of bar found underground" or even, dare I suggest, "Foo is a type of bar found underground". It may require a little discovery but perhaps the 5% can figure it out. Regards, Orange Suede Sofa (talk) 23:57, 9 May 2015 (UTC)

@W. P. Uzer: Thanks for the set of options; it helps make the discussion fruitful. On the topic of readers clicking the link to the IPA help page, some readers don't actually rely on the IPA symbols, or their knowledge of what sounds the symbols represent in the academic arena of phonetics, or even on the help page, but rather on the pronunciation tooltips (example words containing the sound) that pop up when you put your mouse over a symbol. When I say "some users", I'm referring to Quercus solaris, who said so in the section above, but I suspect other readers do the same thing. Therefore, they may not be forming opinions on what phones the IPA symbols represent, but simply relying on the pronunciation of the word they know. Still, would be good to make it clear that the IPA symbols are not used in the "literal" or "accurate" way, so that readers don't make inaccurate comparisons between their variety of English and other languages that use the same symbols. — Eru·tuon 03:35, 7 May 2015 (UTC)

RE: #3, Wikipedia:Pronunciation respelling key suggests "vice" should be respelled VYSS, precisely because VYS would be confusing. I assume the note at the top of this talk page is out-of-date. --2001:388:608C:6902:C57A:4A4F:FE3C:6314 (talk) 06:08, 7 May 2015 (UTC)

Brilliant! It seems the people who design the "respelling" key are just as (not sure which adjective to use here) as the people who design the "IPA" key. Never mind that readers will be totally misled, as long as we can have our fun designing an interesting system. (I suppose it never occurred to them to respell "vice" as "vice"?)W. P. Uzer (talk) 08:06, 9 May 2015 (UTC)

I am confused by your response. The respelling provided by the key for "vice" is entirely non-ambiguous and uses a method consistent with other applications; it does not mislead the readers in any way. Choosing to respell "-ice" as "-ice" in all cases could work in theory, but there is no advantage to using it over the equally clear "-yss", and it is not as easy to generalize. In any case, if you think the respelling system should have more exceptional case respellings (like "-ice"), it would be better to bring it up on that page. --116.240.141.206 (talk) 05:12, 10 May 2015 (UTC)

English speakers do not read "yss" as "ice". They do read "ice" as "ice". No rational respelling system would use "yss" for this (or any other) purpose. W. P. Uzer (talk) 11:12, 11 May 2015 (UTC)

Nobody should even mess around inventing their own English-language respelling systems, because the AHD transcription system (the one used by AHD, even ignoring many "AHD-like" variants) is already excellent, widely known, and just as easy to explain or learn as any respelling (which means, "If you have to consult any key at all, then this key is just as short and sweet as any other"). I think people who build yet-another-respelling-system endlessly pursue some ideal of "even the most clueless users will be able to read a transcription instantly upon encountering it, without even consulting the key, even if they've never encountered the system before." But building one without running into some snags never turns out. If a user is so remedial that they can't be bothered to hover over a symbol and find out "u as in cute", or glance at a key and find out "ī as in wine" or "ĭ as in fit", then the only UI for them regarding pronunciation is to listen to an audio playback. And audio playbacks are wonderful—we all can be glad they exist and we all can use them—but the rest of us like to also have simple transcriptions available as an option, for example, in print, or in situations of muted audio volume. Quercus solaris (talk) 16:26, 16 May 2015 (UTC)

I would just like to add that W. P. Uzer's unstated option 0, keep the status quo, is and will remain my preference. W. P. Uzer has done little but rehash arguments that have taken place on this talk page repeatedly for the last 8 years, and have always resulted in Wikipedia's own diaphonemic scheme being kept. As such, the rationale for my view can be found scattered throughout the archives of this talk page.

Having said this, I would not be opposed to considering whatever changes to the scheme W. P. Uzer would propose that stayed within the framework of a diaphonemic IPA system, though I reject the ones proposed so far. Parentheses for preconsonantal rhotics is confusing and un-necessary, and I do not consider the use of / characters as delimiters to be inappropriate for diaphonemic use. If the consensus were to emerge that WP should abandon slashes, though, my preference would be either for U+2AFD DOUBLE SOLIDUS OPERATOR (⫽) or for distinguishing by typographical distinction like using bold-face type for pronunciation guides, or use of a serif typeface. DavidPKendal (talk) 13:33, 8 May 2015 (UTC)

I'm still not seeing any arguments defending the status quo. How can you (or any of the rest of you) possibly support a system that tells bare-faced lies to readers? I realize you may have argued this elsewhere at another time, but can you at least sum up your position in a few words? W. P. Uzer (talk) 08:01, 9 May 2015 (UTC)

I don't see how Wikipedia's system tells 'lies' of any sort. No, our transcriptions are not phonetically accurate representations of speech which use symbols only in the 'official' way prescribed by the International Phonetic Association; but neither is the phonemic system defined by Gimson which is used in British dictionaries. Would you also propose that we use the symbol 'ɫ' rather than 'l' in milk, for instance?

Wikipedia has different needs to the dictionaries for which Gimson's pronunciation transcription was devised. Gimson's transcription, even with the modifications to it (like the use of 'i' without length mark for the happY vowel) which have since become common, only represents one variety of British English pronunciation. Wikipedia is an international resource, and we cannot justify such provincialism. But since this is an international resource, the international standard for transcription of pronunciations, the IPA, is undoubtedly the most appropriate system to use for this purpose. We are using the IPA in a way that is 'non-standard' only in roughly the same way that the Gimson scheme is non-standard, since the only 'standard' way to use the IPA is phonetically, not phonemically. The only difference is that Gimson's scheme is used by several dictionaries, but there are plenty of IPA phonemic transcription schemes, for English and for other languages, which are only used in one publication or by one publisher (most notably, the incompatible phonemic scheme for British English used in Oxford dictionaries). DavidPKendal (talk) 10:03, 9 May 2015 (UTC)

WP, all transcriptions are to some degree abstractions. Unless you're making a hyper-narrow phonetic transcription of one particular speaker at one moment, you're making some abstractions. Such abstractions are not 'lies'. They are more useful for almost all cases; the trick is to A) keep in mind the limitations of abstraction, and B) tailor which abstractions you're making to the purpose at hand. As can be seen from your reaction to the pronunciation respelling scheme, no transcription scheme can satisfy all the demands made of it.

The primary purpose of English-language pronunciations being given in English-language Wikipedia is for English speakers to know how unfamiliar words are normally pronounced in English. As English speakers typically do not adopt the pronunciation of another dialect of English, giving a pronunciation in various dialects is nearly useless and likely misleading. We give local pronunciations in cases where they differ markedly or interestingly from what is cross-dialectically expected. And in the rare cases where the dialect variation cannot be explained by the well-established systematic correspondences, as in aluminum, we give them each explicitly.

All that said, you clearly have a great deal of knowledge and I look forward to considering specific proposals you could make. — ˈzɪzɨvə (talk) 23:28, 9 May 2015 (UTC)

Of course all transcriptions are abstractions (and lies) to some extent. But that observation does not entitle us to take standard symbols which have standard meanings in the real world (even if those meanings are somewhat variable) and attach our own, wholly non-standard meanings to them. It's simply wrong, just as much as it would be to redefine the kilogram or socialism for Wikipedia's purposes. Use non-standard symbols for these purposes if you must (but there's really no need, as the information we want to convey can be conveyed perfectly well by standard methods), but don't attempt to redefine the standard. W. P. Uzer (talk) 11:12, 11 May 2015 (UTC)

For the sake of argument, let's distinguish here between standards and norms. Standards, like the length of one meter or one inch, are universal: they are usually decided by a committee, and a firm, single definition of them can be obtained by citing one or another source. Norms, on the other hand, are relative: the norm in most of the word is to measure lengths in meters, but if you go to the UK or the US you'll also find measurements in feet and inches. Feet, inches, and meters are all firmly set by standards; the usage of those units in one or another place (ignoring things like the Weights and Measures Act) are norms of those places.

The IPA is a standard. It has an extremely rigorous definition set out in the International Phonetic Association Handbook and each symbol has one, firm meaning. Nobody could doubt that if you used the symbol 'l' when transcribing the word 'milk' as spoken by a typical British or American speaker, you would have made a mistake, because 'ɫ' is the correct symbol for the 'dark L' which most native speakers would use in that word. But then for some speakers, that will also be wrong: those exhibiting 'L vocalization' will use a sound notated 'o' in place of the 'ɫ'. The problem with the firmness of this standard is that, as it is, it only works for transcribing the speech of one particular person at a particular time.

The 'standard' schemes you refer to, by contrast, are mere norms, which employ those symbols in meanings which do deviate slightly from the official meaning, but work around this problem and can apply to large groups of people with broadly similar pronunciations of most words. There is a norm for transcribing Received Pronunciation which is employed by most British dictionaries. All of them are slightly different, but they are mostly the same. This norm is only appropriate for transcribing British pronunciation; though it may also work for some South African and Oceanian speakers, they are not its target audience. There are other norms for transcribing the sounds of American English with the IPA, but since no major US dictionaries use the IPA there are more of them. It doesn't mean the dictionaries are wrong for not using the same norm.

Similarly, it doesn't seem wrong to me that Wikipedia should establish its own norms for using the IPA. The existing norms don't suit our purposes, because they apply to groups of people which are too specific to be useful for us, so we need another one. An analogy is physicists using Kelvin because Celsius has an arbitrary value at 0°, whereas 0 Kelvin is a universal constant. They simply have another norm. DavidPKendal (talk) 18:41, 11 May 2015 (UTC)

On top of that distinction, the analogy with the definition of socialism is quite apples/oranges: telling people an incorrect definition of socialism doesn't normally have an educational value (although, see lie-to-children) whereas having this system be diaphonemic does have educational value as long as it efficiently informs readers of the intended pronunciation in their own dialect. Deviating a little from typical norms for the IPA by using it to represent a diasystem may have some theoretical issues (as has been discussed) but as we are concerned more with practical issues, it should be fully open to us. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 18:58, 11 May 2015 (UTC)

As should be clear by now, it is practical issues that I am concerned with. The system does not efficiently inform readers of anything, since it uses conventions that they cannot possibly know. And worse than that, it re-uses symbols that they very likely do know, so they may have no reason whatever to think that they need to click and read guides and etc. That is the fundamental problem as I see it. W. P. Uzer (talk) 08:31, 16 May 2015 (UTC)

A concern with practicality in this context means a concern with mispronunciations. Much of what you've brought up as problems wouldn't actually lead to mispronunciations. So far, the only actual mispronunciation you've demonstrated as even possible is one from representing the BATH vowel as ⟨aː⟩, which might prompt some readers to mistake it for ⟨ɑː⟩. But you were complaining about the system even before we adopted this practice, going so far as to explicitly complain about having two transcriptions for the BATH vowel. If there aren't any other mispronunciations coming from the system using IPA symbols diaphonemically (and you can take this as a request for more of such examples), then we do indeed efficiently convey the intended pronunciation, even for IPA-literate readers who have mistakenly understood our system to be closer to typical conventions of IPA transcription. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 09:21, 16 May 2015 (UTC)

Well, analogously to the BATH vowel, there is the CLOTH vowel, which is shown as long on Wikipedia's system even though many speakers have it short, and might consequently be pronounced as a long o or long a by people who ought to be pronouncing it as a short o. I don't know how GAm speakers might react to our symbol for the LOT vowel, but I don't see any particular reason why they should look at it and think long a. But besides this, I absolutely reject and condemn the attitude that the misinformation we provide doesn't matter if it only affects foreigners, or readers who have the impertinence to take an interest in varieties of speech other than their own. This is surely totally at variance with Wikipedia's policies and guiding principles, and I don't think people here really think that either; it's just a desperate attempt to find arguments in support of a system that you've all got used to using, but is demonstrably not the right one for Wikipedia. W. P. Uzer (talk) 13:29, 16 May 2015 (UTC)

I'm having a little trouble visualizing the problems of the CLOTH vowel. With the father-bother merger, wouldn't GA speakers typically pronounce the CLOTH vowel as usually [ɑː]? Maybe International Phonetic Alphabet chart for English dialects can provide you with some guidance in being more explicit.

Regarding the issues of non-native speakers and those interested in other dialects, you are bordering on demagoguery. Every time you bring this up, you've ignored my counterarguments and then gone on to repeat it.

You first brought up the point that the diaphonemic system doesn't help amateur dialectologists on April 18, though I pointed out that the system doesn't address that interest. You repeated your argument again on April 23 with no further refinement, as if repeating your argument again would somehow make it more persuasive. You again brought it up on May 6, characterizing it as an untoward restriction and I (along with Xyzzyva) pointed out that this isn't the kind of information dictionaries attempt to address in the first place. You didn't respond.

You brought up a tangential claim on April 26 that readers will glean dialectological information from our system; when I argued that this assumption was unreasonable on your part, you simply didn't respond. When I said that incorporating dialectical information shouldn't come at the expense of efficiently indicating pronunciation, you simply said "I don't accept that."

When you implied on May 4 that this lack of focus on dialectological information in this system somehow prevented users from indicating dialectological information at all, I pointed out that there's nothing against indicating this information alongside the diaphonemic transcription. Yet here you are again, with your sarcastic tone, making that same point as if I hadn't already rebutted it.

You've also repeatedly brought up the issue of non-native speakers as an anti-foreigner bias; forgive me if I sound rude, but as it's my understanding that the acquisition of English as a second language requires choosing a dialect early on (typically either RP or GA;), you sound like you don't know what you're talking about each time you characterize it this way. If you know something I don't (which is possible, since I'm not particularly knowledgeable about second-language acquisition) please enlighten me; but I'm under the impression that the working knowledge of English necessary to read at the English Wikipedia would normally involve having "chosen" a dialect already. — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 17:33, 16 May 2015 (UTC)

Æµ§œš¹: As I understand it, speakers in the United States who distinguish /ɔ/ and /ɑ/ pronounce CLOTH with /ɔ/. This is because of vowel lengthening before voiceless fricatives (the lot–cloth split). So CLOTH had a different vowel from LOT, and didn't undergo the father–bother merger. CLOTH is only pronounced with /ɑ/ by speakers who merge /ɔ/ with /ɑ/ (cot–caught merger). — Eru·tuon 19:57, 16 May 2015 (UTC)

OK, so WP's argument is that our transcription of the cloth vowel as ⟨ɒː⟩ might prompt readers without the LOT-CLOTH split to pronounce /ɔː/ or /ɑː/ instead of /ɒ/ and speakers with such a split to potentially read ⟨ɒː⟩ as /ɑː/. Is that right? — Æµ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 08:02, 17 May 2015 (UTC)

layer example?

Latest comment: 8 years ago8 comments5 people in discussion

since when were layer as in "one who lays" and layer as in "dressing in layers" pronounced differently? -Cookie Fonster^talk_sign! — Preceding undated comment added 21:50, 12 March 2015 (UTC)

I suspect it's one of those things like the horse–hoarse merger and the pour–poor merger—for example, north, born, war, force, more, boar, boor, moor, tourist all sound the same in my accent, but this page preserves distinctions among /ɔːr/, /oʊr/, and /uːr/ because for other people with other accents, it matters. Quercus solaris (talk) 22:35, 12 March 2015 (UTC)

Speaking of which, when I write transcriptions in WP articles, I have to stop and think about ɜr, ʌr, and ʊr, because "they's all the same where I come from". Thus nurse, word, girl, fern, furry, borough, hurry, and even courier are all as one to me. Guess I could feel deprived, but I kind of like the simplicity of it. Quercus solaris (talk) 22:41, 12 March 2015 (UTC)

They are not pronounced differently in standard British English (/ˈleɪə(r)/ for both), but some dialects pronounce the stratum sense like lair (/lɛə(r)/), hence the clarification of one who lays. Would it be better to reduce this to a footnote, since most speakers will not make the distinction? Dbfirs 08:34, 13 March 2015 (UTC)

Per the OED, both meanings are pronounced with two syllables. — kwami (talk) 22:48, 13 March 2015 (UTC)

Yes, that's what I said (/ˈleɪə(r)/ for both in the OED), but not all dialects use standard English. Wiktionary also has /lɛə/ for layer, and some dialects have just /lɛ:/. Similarly, lair can be either one or two syllables (/lɛə(ɹ)/ or (/lɛ:(ɹ)/). (Dialects also vary in the interpretation of /(ɹ)/, of course.) Dbfirs 22:57, 13 March 2015 (UTC)

There is no "standard English". — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 09:56, 16 June 2015 (UTC)

I'd keep it. "Layer" as in stratum is the same as "lair" for me (leaning almost toward two syllables but not quite), but unlike "layer" as in "one who lays" (which is firmly two syllables, almost to the point of being two words, "lay" and "er", as it sounds when we add the comparative -er to a word that doesn't normally have it and ends in vowel, like "macho-er". Maybe just find a better example? — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 09:56, 16 June 2015 (UTC)