Help talk:IPA/English/Archive 1

Purpose
This page is meant to be the key for the IPA-en template (which displays "") that is used as a broad pronunciation guide to key words in Wikipedia articles. It is not meant for phonetic detail, dialectical differences, or non-English phonologies. Please keep it as simple and accessible as possible, as many of our readers are not familiar with the IPA. kwami 19:41, 15 October 2007 (UTC)

As for syllabification, that is not distinctive in English and does not need to be indicated. Showing syllable breaks just sparks edit wars with people who think they should be somewhere else. Problem is, English has ambisyllabic consonants, which cannot be represented by the IPA. kwami 23:03, 15 October 2007 (UTC)

Problems
The obvious problem is that there is no single pronunciation of English. Attempts to create a standard either privilege one dialect over others; create an artificial pronunciation no one speaks; or are a confusing mixture of both. This particular mode picks up some odd bits of phonetic trivia (like the insertion of a schwa before a syllable-final r, which does not occur in many varieties of English) but omits other pan-English phonetic developments, like voiceless stop aspiration - which, of course, cannot be determined by rule without reference to the supposedly "non-distinctive" syllabification. RandomCritic
 * This is a pan-phonemic transcription, and so doesn't suffer from these problems. Aspiration is irrelevant, and the schwas are not relevant to those who don't pronounce them. kwami
 * Then why are you including the one and not the other? Because they happen to occur in your personal speech variant? RandomCritic 17:22, 16 October 2007 (UTC)
 * Actually, I have the aspiration and not the schwas, so it's the exact opposite of my dialect. But aspiration is allophonic, and so has no business in a phonemic description. For example, if we were to write tie and die, how would we write sty? It would have to be either  or , because there is no third phonemic stop, and therefore  is not an option. Since neither of those would be acceptable to our readers or editors, we can't be phonemic and also indicate aspiration. The schwas before ar, on the other hand, are phonemic in most English dialects, including RP, and therefore should be including even if you and I do not have them. kwami
 * You really don't grasp that the schwas are conditioned variants? Do I need to spell it out for you? RandomCritic 12:35, 17 October 2007 (UTC)
 * I guess you do. Could you explain how Sirius /ˈsɪriəs/ and serious /ˈsɪəriəs/ are conditioned variants? They look like minimal pairs to me, and that is how the OED treats them. kwami 18:32, 17 October 2007 (UTC)
 * Wow. Just wow.  I really missed this fantastic piece of dullness.  What a shame!  For the rest of you, Kagami failed to grasp that he needed to provide a minimal pair between [iːɹ] and [ɪəɹ], not [ɪɹ] and [ɪəɹ].
 * Phonology. It's a science!  You could study it!RandomCritic (talk) 13:00, 6 April 2008 (UTC)

Not surprisingly, IPA transcriptions of English words used in America, England, and Australia are all different -- and slightly more surprisingly, phoneticians in these countries all use the IPA in slightly different ways, rather de-internationalizing it, and making it very difficult for a person trained in the tradition of one country to read another person's transcriptions correctly. Even more problematically, "traditional" or "standard" transcriptions of sounds may be read in ways that are quite misleading from the point of view of IPA, especially its canonical vowels. A Midwesterner seeing [ ɒ ] in [ stɒp ] may suppose that it is the same sound he or she uses in pronouncing "stop", though in fact the latter is closer to [ ɑ ], or even [ a ].
 * Again, these are [phonetic] details not relevant to a /phonemic/ transcription.

It is likewise "traditional" to transcribe the "long a" and "long o" diphthongs as [ eɪ ] and [ oʊ ]. But in America, England, and Australia alike, the dominant pronunciations have much lower nuclear vowels than [ e ] and [ o ], while the off-glides are closer to [ i ] and [ u ] than [ ɪ ] and [ ʊ ] in most pronunciations, outside of the now vanishingly rare RP. The "traditional transcription" thus enjoins a pronunciation that almost no English speaker actually uses.
 * Again, irrelevant detail. /oʊ/ is the vowel of bone, however you pronounce it in your dialect. After all, the only point of this key is to show you how to pronounce a word in your dialect. If you're interested in someone else's dialect, this key is obviously of no use, because of the objections you raise.

The scheme shown assumes a rhotic dialect, but the vowels are for the most part chosen from a (typically non-rhotic) British variant of English, strongly influenced by RP. This cannot be back-translated into a non-rhotic dialect without the use of a rule involving deletion of a non-onset r and compensatory lengthening; a task considerably complicated when syllabification is considered irrelevant!
 * Give me an example that is problematic. If you speak a non-rhotic dialect, just drop the ars and keep the length, as in the key words. If you merge some of the vowels, then go ahead and merge them as in the key words. No problem that I can see. If there are problems, that only means we missed some phonemic distinctions.

I also note the -- to me, quite novel -- use of [ ɨ ] to represent the slightly lower and more centralized variant of [ ɪ ] found in unstressed vowels. The symbol [ ɨ ] canonically represents a quite different-sounding sound, the desired sound being not nearly so high, nor so central -- being, indeed, closer to [ ɪ ] than to any other canonical IPA vowel, though of course not identical; while a fully stressed [ ɪ ] (perhaps only heard in very emphatic speech) is more front than the canonical [ ɪ ].
 * We're not using [ɨ], we're using /ɨ/ - quite a different thing. We could've picked /♠/, but that would have thrown people for a loop. There is no good symbol in the IPA for this, and /ɨ/ is generally used by phoneticians who wish to keep it distinct from /ə/. This is covered in some of the links at the bottom of the page.
 * Who is "we", anyway? User:Kwamikagami and who else?RandomCritic 17:21, 16 October 2007 (UTC)
 * The people who wrote the IPA chart for English and International Phonetic Alphabet for English, where three reduced vowels, using (for some dialects) the same symbols, are distinguished. kwami

In the end, I am not particularly pleased with the use of IPA to represent some pan-English, not-quite-phonetic, yet not wholly phonemic transcription. I would rather see either: (1) consistent, side-by-side uses of a few common dialects;
 * - that would be quite a mess, and a real editorial problem
 * Not nearly so much of a mess as this one. RandomCritic 17:21, 16 October 2007 (UTC)

or (2) one dialect, of any type, used consistently, and identified, and accepted as WP standard;
 * - not politically feasible; no matter which dialect we choose, we'd be accused of cultural imperialism

or (3) an abandonment of IPA altogether in favor of a phonemic scheme that can be translated by the aid of certain rules to an approximation of one of several dialects of English.
 * - you can see that here: Help:Pronunciation respelling key, but what an outcry if you try to use it! I've tried explaining that such a convention might be preferable because the IPA is often misinterpreted as representing specific sounds, but people really get upset with pronunciation respellings. There's no real reason to reject the IPA; as far as I can tell, the distinction of [] and // takes care of your objections.
 * No, it doesn't, because it's based on a gross misunderstanding of the use of those symbols. RandomCritic 17:21, 16 October 2007 (UTC)
 * It requires an understanding of the difference between phonetic and phonemic use of the IPA. The Association itself, in its Handbook, uses  as, for example - a phonemic transcription does not have to be phonetically accurate, and in general it cannot be, not if there is any complexity to the system. kwami

Of course this should be something that has some precedent of usage, and is not something made up by a WP user yesterday. RandomCritic 05:29, 16 October 2007 (UTC)
 * This is pretty much standard, and have been in the Talk pages for years. kwami 06:54, 16 October 2007 (UTC)
 * As I pointed out right at the top, this is not a true phonemic scheme but includes all sorts of (dialectical!) phonetic details.
 * No, they are phonemic distinctions. kwami
 * It is moreover a wrong and a misguided assumption that all English dialects have the same underlying (phonemic) representations.
 * It does not assume that; it's a pronunciation key that will enable readers to pronounce words in their dialect by analogy with the key words in the chart. Theoretical conclusions based on that are misplaced. kwami
 * In fact they do not: almost every dialect will have a different underlying representation appropriate to its own system.
 * True, but many English dialects are close enough that this is not a fatal problem. (Trying to include Scots or Indian English might be asking too much, though.)
 * Phonological analysis is a synchronic form of analysis that simplifies uniform variations within a single spoken language variant; it has nothing to do with diachronic or comparative analyses. Creating a system of pre-dialectal or pan-dialectal representation that can (with a knowledge of the proper rules) be converted into a dialectical pronunciation is something entirely different from producing a phonemic representation.
 * Correct. kwami
 * I have repeatedly noticed a failure on the part of WP editors to understand what a phonemic representation is -- or indeed, what phonology is, which is what, I suppose, produces nonsensical statements like "/oʊ/ is the vowel of bone,". Phonology does not determine underlying forms based on lexicographic fiat, but upon an analysis of sounds actually occurring in the spoken language -- and if [ oʊ ] is not an allophone that appears in a given spoken variant, then it is never going to appear in the phonemic representation. (And from a pan-dialectal or pre-dialectal point of view, the vowel is probably [ oː] anyway.)
 * Incorrect. There is no pan-dialectal phonetic transcription, and the symbols used in a phonemic analysis are irrelevant. True, an "elsewhere" allophone is generally chosen, but that is for the convenience of the reader, not a theoretical requirement. I forget now the phonologist who chose symbols something like /♠/, /♣/, /♥/, /♦/, to represent the four vowels of the Micronesian language he was working on, in order to make precisely that point. What is important here for us is that the symbols be intuitive and accessible to our readers, not that they carry a specific phonetic value. kwami
 * Using the word "intuitive" wrt IPA symbols is silly -- people who are not intimately familiar with IPA most frequently complain that it's not intuitive. There's nothing "intuitive" about either the symbols themselves, or the specific values being used here; what is particularly striking is their lack of connection with the facts of any language.
 * Of course the IPA is intuitive. Much more intuitive than, say, using numbers would be; it just takes a little getting used to. kwami
 * My point, apparently missed, is that pan-English doesn't exist -- so there can be no "phonemic" representation of it.
 * No, I didn't missed that point, but again it's not relevant. This is not a theoretical description, it's a key, and a much more helpful one than the often unintelligible mess that's out there.
 * And applying phonetic symbols with a specific value to create glyphs that are never pronounced that way is neither doing phonology nor producing a useful Help scheme. I'd much rather see your card suit symbols than a set of phonetic representations which are necessarily going to be either misread or misapplied. RandomCritic 12:35, 17 October 2007 (UTC)
 * Again, you seem to have the impression that the IPA can only be used as a phonetic alphabet. I don't know where you get that idea from.


 * In any case neither the phonemic representation (of any dialect) nor some pan-English scheme is going to be worth anything without supplying the rules by which that scheme can be converted into something that can actually be pronounced, i.e., a phonetic representation. RandomCritic 17:21, 16 October 2007 (UTC)
 * That scheme is supplied: the chart gives sample words for each symbol. All you have to do is pronounce the symbol the way you pronounce that word, and you have the correct pronunciation in your dialect, or at least you do if your dialect is RP, GA, Oz, and some others that fit. In other words, this is no different than the pronunciation guide of, say, Webster's 3rd, which is also inter-dialectal, except that it uses the IPA instead of some local in-house convention. Can you give me any examples of where this system breaks down for the national dialects? So far you criticisms have been theoretical, without practical examples, and this is after all a practical situation: a pronunciation key, in Help space, not a theoretical treatise, which we already have in article namespace. kwami 21:49, 16 October 2007 (UTC)
 * That is exactly not how the IPA is to be used -- it's not a series of random glyphs, and when people who actually know the IPA read it they are not going to read it in this sense of a unique (and probably OR) scheme that only exists in the pages of WP -- they are going to read the symbols with the values the symbols actually have.
 * So, is the Handbook wrong to use  to represent [tʃ]? Or any of the litterally dozens of other cases where for convenience they use an IPA symbol for a value far removed from its defined phonetic value? kwami
 * At which point it becomes apparent that something quite bizarre is being presented -- a "phonemic" transcription of a non-existent dialect of English, cobbled together from a bunch of half-understood pronunciation charts (with no actual phonological analysis having been done).RandomCritic 12:35, 17 October 2007 (UTC)
 * Am I supposed to be doing OR phonological analysis here? And how is this bizarre to a user of, say, Webster's 3rd, which also uses an inter-dialectal transcription? Your objection seems to be that the IPA shouldn't be used for such things; can you tell us what should? Because the folks around here are not goind to be happy with a pronunciation respelling using anything but the IPA. kwami
 * P.S. You put a flag on the help page stating that the "factual accuracy of this article is disputed". Which facts? This is a convention, a help guide, not a statement of truth. Please give an example. kwami 21:52, 16 October 2007 (UTC)
 * Okay, you disagree with this approach philosophically, as a matter of personal opinion. You still haven't come up with anything factually wrong about it. kwami 18:32, 17 October 2007 (UTC)

Bottom line
The bottom line is that it is extremely annoying that User:Kwamikagami, though he doesn't seem to understand either the theoretical basis or the facts behind various transcriptions of English (much less the principles behind the English pronunciation of classical names, but that a whole different can of worms) feels the necessity to go hither and yon around WP imposing his misunderstandings upon hundreds of articles and thousands of pronunciations, creating errors -- often highly ridiculous and grotesque ones -- which may not be discovered and corrected for months or years. Why doesn't he just leave it alone? RandomCritic 12:35, 17 October 2007 (UTC)
 * Again, please give examples of these "grotesque" errors. You need to back up your claims. As for the principles behind Classical names, I wouldn't be at all surprised that I don't understand vowels that don't exist in my dialect, but that's a side issue to this page. I don't leave it alone because there are contradictory standards of the IPA used here in Wikipedia, so that often the result is ambiguous. We need some standard so that e.g. we know which sound /e/ is supposed to represent. kwami 18:32, 17 October 2007 (UTC)
 * The pronunciation key presented here, looks to me like an excellent approach to establishing a broad phonemic transcription of English, making balanced compromise choices, not too far from most dialects and sticking to broad interpretation guidelines of the IPA. It is customary in phonemic transcriptions to select conventional symbols from the phonetic symbol tables, that are somewhere near the real sound in the continuous phone space, without using the more unusual or heavily marked up symbols. For example, most applications of IPA to English use /r/ (a trill), whereas the sound this symbol transcribes is rarely realised that way. I am convinced that anyone pronouncing these transcriptions with exact phonetic values would be instantly onderstood. &minus;Woodstone 20:01, 17 October 2007 (UTC)
 * Just came across Help:Pronunciation, where it says "The goal is that phonemic interpretations should not differ depending on the reader's regional dialect." That means Wikipedia editors have already decided to follow the approach that RandomCritic rejects. kwami 00:52, 18 October 2007 (UTC)

I'm not an expert, but it all looks pretty good to me. Thanks for all the work it took to assemble this. Since it is widely linked, we'll see what general readers make of it in a little while. —Michael Z. 2007-10-19 21:40 Z 


 * Since the tag says the facts of the chart, not its appropriateness, are in dispute, and RC and TI have not provided any facts which they dispute despite several requests, I'm removing the tag. kwami 05:59, 21 October 2007 (UTC)


 * For the record, I have never been requested to provide facts I dispute nor have I ever said anything about disputing facts. Timeineurope 07:54, 21 October 2007 (UTC)
 * Sorry, my bad. It was just RandomCritic. kwami 01:51, 22 October 2007 (UTC)

Ok, here's a problem I'm having here. This pronunciation table does not help. Whatever happened to the dictionaries I used as a kid that had pronunciations in mostly roman letters, with the occasional funky lines or dots over vowels? Does Webster have a patent on those? Bottom line is you're reinventing the wheel, and doing a piss poor job.71.185.131.144 (talk) 04:51, 1 March 2008 (UTC)


 * Solution: You need to learn how to use another dictionary. We can't give Webster's transcription for you, Random House's for someone else, American Heritage's for someone else, the OED's for yet someone else ... kwami (talk) 06:55, 1 March 2008 (UTC)

Martian pronunciations
Among the absurd pronunciations, used by nobody in the world, that User:Kwamikagami has come up with are:



These are ridiculous, and show that the user hasn't the slightest idea of what he is doing -- yet bulls ahead anyway. I daresay there are more such out there waiting to be discovered. RandomCritic 15:32, 18 October 2007 (UTC)
 * Rather than trying to stir up conflict, why don't you contribute something, such as saying (a) what is wrong, and (b) what would be better? Not a difficult thing to do. kwami 23:39, 18 October 2007 (UTC)
 * I apologize for never returning to this. I was unfortunately under the misapprehension that no one could possibly not see the problems: (1) false length, false diphthong, false symbol; (2) false quality, false quality (and length!), false voice, false symbol; (3) false diphthong.  When you use IPA as routinely as I do, you perceive the false notes instinctively.  I had forgotten that to some people it's a kind of hieroglyphic which is encoded and decoded by eye alone, without regard to actual sound values. RandomCritic (talk) 12:21, 6 April 2008 (UTC)

Phonetic vs. phonemic
Several editors object to using phonemic transcriptions for English words, or non-regional transcriptions, or perhaps both. However, this doesn't have to be an either-or choice. For place and personal names, there's the internationally recognized form, and often a differing local form. With Toronto, for example, a non-regional transcription would be something like, and the local pronunciation pronunciation could be transcribed something like , assessing a different subset of the IPA chart. That is what is suggested at Help:Pronunciation. Before we get into any more edit wars, let's discuss if this is how we want to go with Wikipedia. The alternative, as I see it, is to list London, Sydney, New York, Atlanta, Dublin, Johannesburg, Aukland, or Los Angeles as well as Toronto pronunciations of "Toronto", all phonetic, in an attempt to avoid theoretical complications. kwami 23:56, 18 October 2007 (UTC)


 * Support: as Kwami suggests, it seems to make sense that for names and words which have a commonly-used and internationally-recognized pronunciation, that pronunciation should be transcribed using IPA. However, if there are common local pronunciations, those may be added as well. For instance, the way "Beijing" is pronounced in English is not the same as how it is said in Mandarin. Cheers, Jacklee 00:44, 19 October 2007 (UTC)

If I understand correctly, the opposing side feels that we shouldn't create abstract pronunciations that don't exist in the real world. At least one person objects to any phonemic transcription because he rejects the concept of the phoneme alltogether. Here's an simple example:

Collet is, per the OED (RP), /ˈkɒlɪt/, and per W3 (GA) /ˈkɑlɨt/. Now, do we really want multiple transcriptions for a simple word like collet? (Aussie might be the same as RP for this word, but it won't be for others, meaning yet a third national standard. And then there's SA, Ireland, and other countries. Currently we only list GA, which some RP speakers might rightfully object too.) We can combine the two, per the chart on this page, for */ˈkɒlɨt/. (Asterisk for a constructed form.) Now, RP speakers know (or will know, once they check the chart) that for them there is no difference between /ɨ/ and /ɪ/, while GA speakers know (or will soon realize) that for them there is no difference between /ɒ/ and /ɑ/. Thus speakers of either standard can read the transcription correctly according to their dialect. This is precisely the tack that W3 takes for various US dialects, though of course they use their own in-house symbols, which most Wikipedians fiercely reject. (Believe me, I've tried that too. And this isn't OR, because W3 symbols when read broadly are exactly equivalent to our */ˈkɒlɨt/)

If we go further, and reject a phonemic approach entirely, then the number of transcriptions multiplies, if only because GA isn't very well standardized, and the single transcription [ˈkʰɑˑlɨt] will not be sufficient for the entire country. kwami 01:33, 19 October 2007 (UTC)


 * I speak as someone who has not studied phonetics and whose acquaintance with the IPA is from the pronunciation key of the Oxford English Dictionary. I'm not sure what you mean by W3 (GA), but I believe what most ordinary readers want is simply to have some reasonable manner to know how to pronounce unfamiliar terms. If that is what taking a phonemic rather than a phonetic approach, then I am all for it. Isn't that the approach that major dictionaries and encyclopedias take?
 * I would suggest that a distinction be drawn between "ordinary" English words (such as collet) on the one hand, and non-English terms which have not been fully assimilated into English and foreign terms (such as personal and place names) on the other. Perhaps a phonemic approach can be taken for the former and a phonetic one for the latter. What I mean is that for the latter category, the way the term is pronounced in its native language should be set out, and if there is a different but commonly-used way in which it is pronounced in English that should be given too. Cheers, Jacklee 02:00, 19 October 2007 (UTC)

W3 is Webster's Third International Dictionary, and GA is "general American" (newscaster standard).

Another approach to this which I forgot to mention, and which TI has switched to and I reverted pending discussion here, is to use parentheses for sounds which differ between dialects. So, for example, our /ɪər/, which could be read [ɪr] or [ɪə] or [ɪər], depending on dialect, would become /ɪ(ə)(r)/. Personally I find it more difficult to read, and I suspect it will cause just as much confusion among our readers. kwami 03:54, 19 October 2007 (UTC)


 * Let's compare the virtue of transcribing Charles as /ˈtʃɑrlz/ and transcribing it as /ˈtʃɑː(r)lz/. The first transcription leaves anyone who doesn't follow the link to this article thinking that there is always an r sound in that word. The second transcription leaves them thinking that Charles can be pronounced both with and without an r sound. Only the latter is true. Timeineurope 10:38, 19 October 2007 (UTC)
 * But since nearly all the 20,000 Wikipedia articles with pronunciation guides give a single transcription, without parentheses, changing this chart will have no affect on those people. In fact, if they do come to this page, your method will falsely confirm that Charles can only be pronounced with (or without) an ar. (Assuming it's a typical article without parentheses.) kwami 11:20, 19 October 2007 (UTC)
 * Obviously a change to this chart would have to be followed up by changing a lot of articles (just like you have followed up the creation of this chart by changing a lot of articles). Timeineurope 12:22, 19 October 2007 (UTC)


 * Here's a sample I found of the difference:


 * Kraken ( kra’ ken | IPA RP: | GA:  | AuE: ) are legendary sea monsters of gargantuan size ...


 * or:


 * Kraken ( kra’ ken, ) are legendary sea monsters of gargantuan size ...


 * kwami 12:10, 19 October 2007 (UTC)

"Bath" words and "cloth" words
AFAICT this scheme has no unified way of transcribing "bath" words (words that have or its equivalent in RP and AuE but /æ/ in GenAm) and "cloth" words (words that have  or its equivalent in RP and AuE but  in GenAm. —Angr 19:05, 19 October 2007 (UTC)
 * Such words would need to be given two transcriptions, like vase [veɪs, vɑːz]. Unless you can think of something? kwami 21:09, 19 October 2007 (UTC)
 * Would this be a problem in practice? Wikipedia is not a dictionary, so IPA transcriptions only have to be provided for articles with terms in their titles that readers might have difficulty pronouncing, not every article. Cheers, Jacklee 23:23, 19 October 2007 (UTC)

Rhotic vowel formatting
When we doubled the number of rhotic vowels (or vowel-ar sequences, many of which I do not control), in the draught before this page was created, it was suggested that all pre-rhotic vowels should be lax. We followed the OED in using schwa for the distinction, as follows:

This is what we now have in the chart. One advantage of this system, besides conformity to the OED, is that people like me can simply ignore the schwas, making it easy to read. However, in going over actual practice in Wikipedia, what I often see instead is to use the existing free vowels plus ar:

The advantage of this format is that it mirrors the non-rhotic vowels (that is, it's a broader transcription), meaning there is only half as much to learn for someone not familiar with the IPA; it is also extremely easy to read for someone like me who doesn't make all these distinctions. The more I think about it, and the more articles I see on Wikipedia, the better the second format looks.

What do you all think? There's still time to switch over. kwami 06:16, 21 October 2007 (UTC)


 * At first impression it looks good, but I have some doubts about /eɪr/ in bear and /oʊr/ in four. Those are pretty far from their phonetic values. How about /ær/ and /ɔːr/? Mary could then have /æːr/. The list is less systematic then the current choice. No firm opinion yet. &minus; —Preceding unsigned comment added by Woodstone (talk • contribs) 10:00, 21 October 2007 (UTC)


 * /æːr/ and /ɔːr/ would introduce length distinctions which are currently redundant, increasing the likelihood of misreadings. Also, /ɔːr/ would be ambiguous, as it's already used by different people for both /ɔr/ and /ɔər/. How far off /eɪr/ and /oʊr/ are phonetically depends on dialect; for a lot of Wikipedians they're pretty close, or so it would seem from their preferences.


 * Let's try to keep a minimum of readability here. It's bad enough having to go from to  – are we really going to require readers to go from  to ? That will just lead to even more misunderstandings by people who don't click the link leading here than the present system already does.


 * I continue to fail to see what's so awful about putting the r's in parentheses, ie instead of . The first alternative will be understood by everyone to mean that the word can be pronounced either with or without the . The second alternative will be misunderstood by many readers to mean that the word is pronounced with an, period.


 * Timeineurope 11:31, 21 October 2007 (UTC)


 * Okay, it seems that /eɪr/ and /oʊr/ would be counter-intuitive for non-rhotic speakers. Maybe the people using them here are rhotic.


 * Personally, I find the parentheses distracting to read, but more seriously, that option is hardly found in the 20,000 Wikipedia articles which use the IPA, which will make it seem even more like the ars are obligatory when the parentheses are left out. If we don't give an option in the chart and just leave out the ars, rhotic speakers are going to recognize from the spelling that they should say them, and if we leave in the ars, non-rhotic speakers are going to drop them, because it's automatic for them to do so. It won't matter, of course, if we use parentheses in the articles, but if we have them here in the chart, people will think the ars are required in all those articles which lack parentheses. kwami 12:15, 21 October 2007 (UTC)

We should stick to the principal of understandability and simplicity for readers. That said, neither version bothers me—I know what a schwa is from elementary school, but I think the length mark (ː) is basic IPA used in many places (and ought to be added to the key).

Let's not worry too much about what is in the 20,000 articles with IPA. Many of them will be non-English or general IPA about phonetics/phonemics, and many of them will simply be inconsistently formulated. This new key should be seen as a recommendation, but also as general enough to help a reader figure out most of the English IPA seen in the wild.

The optional r's in parentheses do add a bit of visual clutter, but they are easy to understand, and make the assumption explicit for readers. I don't see a problem with using them or not. —Michael Z. 2007-10-22 18:12 Z 


 * It seems to me that if these pronunciations are supposed to be phonemic, they are overly specific. Is it really necessary to specify either the schwa or the lengthening? Doesn’t the /r/ imply a schwa in non-rhotic dialects? Doesn’t the base vowel imply lengthening or diphthongization (in some dialects)? How about using simply, , , , , , , , , , , and ? Perhaps that might lead to objections that and  look too much like something that would be pronounced . If so, how about using an r modifier (Unicode 2B3) or a rhotic hook (Unicode 2DE) instead of “r”? Thus: , , , , , , , , , , , and . Or: , , , , , , , , , , , and.
 * It seems to me that putting r’s in parentheses is ugly and unnecessary. The objection that non-rhotic speakers will be confused by a plain /r/ underestimates their intelligence; after all they have no problem with r’s in plain text. I suggested above using an r modifier or rhotic hook; would that be an acceptable substitute for an r’s in parentheses?
 * By the same token for non-r-colored vowels, is it necessary to specify the lengthening of, , , , and and the diphthongization of  and ? Don’t the base vowels imply lengthening or diphthongization? --teb728 10:21, 23 October 2007 (UTC)


 * Implementing any of your suggestions would decrease the readability of the transcriptions.


 * Anyone without the necessary level of knowledge about English pronunciation would conclude from the transcription that that name is pronounced with an, period.


 * Timeineurope 17:59, 23 October 2007 (UTC)


 * The feedback we've gotten from non-rhotic speakers suggests they find the /ir, er, or/ transcription counter-intuitive. I like it myself, but then I'm rhotic. However, a little redundancy is helpful; the way we have it now, if someone writes /er/, we know they're not following the chart. kwami 18:17, 23 October 2007 (UTC)

/ɪ/ vs. reduced / ɪ /
One RP editor objected to using <ɨ> for the reduced vowel of roses, saying it should be <ɪ>. However, it appears that the OED itself has switched over to this usage, though using a more accurate symbol <ᵻ>. For example, the OED lists a "British" pronunciation of parallelepiped as, clearly distinguishing /ᵻ/ from both the other reduced vowel, /ə/, and the full vowel /ɪ/. kwami 00:09, 22 October 2007 (UTC)


 * The /ᵻ/ (U+1D7B, Latin small capital letter i with stroke) appears to be absent from every default Mac OS X font, as well as from Code 2000. It appears as a blank square in Safari, and as a question mark in Firefox (doesn't even look like a problem there)—first IPA character I've ever seen fail on my machine.


 * I suggest we don't use it for practical reasons. As near as I can tell, the OED 2nd and online editions all use the small capital ɪ. —Michael Z. 2007-10-22 16:37 Z 


 * Yes, it's not working, so I'll remove it. I propose manually striking out ɪ like this: . If people don't bother, what we're left with is the old OED format, which is good enough in most cases. (Though occasionally it is contrastive.) It's not actually IPA, but an unofficial extension that's been in use in the sidelines for years and just recently adopted by the OED. (Yes, the online OED does use it now, at least in the parallelepiped entry.) Official IPA would be to use ɪ with a diaresis, which gets a bit clunky. kwami 16:53, 22 October 2007 (UTC)


 * The OED New Edition pronunciation guide explicitly says that their / ɪ / symbol represents free variation between /ɪ/ and /ə/, so I think it's more that they're trying to represent varying pronunciations than that they've abandoned the traditional /ɪ/ analysis.  As it's an innovation, it'll only be found in the New Edition entries, which are mostly words beginning with letters from M to P.--JHJ 16:41, 23 October 2007 (UTC)
 * That's what the Webster's ə-dot represents. Note also they've gone from /ɪ/ to /i/ for the end of happy. kwami 17:43, 23 October 2007 (UTC)
 * Given what the OED actually says about its system, I'm not too keen on the relevant bit at Unstressed and reduced vowels in English which seems to make the same misunderstanding. Do you actually have a source that justifies distinguishing /ɪ/ and "schwi" at the phonemic level in non-American English (and even in American English for that matter)?--JHJ 19:57, 23 October 2007 (UTC)
 * I don't have any sources with me. I'd have to check Ladefoged. But I can demonstrate (OR) the difference in GA: the i in battleship is unstressed but a full vowel. To me at least it would be wrong to pronounce it with the e of roses, yet in context (a battleship fleet) it's clear that only the a is stressed, at least in my pronunciation. Of course, compound words behave somewhat differently than non-compounds, so there's not likely to be much confusion, and using the typical dictionary convention of marking unstressed ship with secondary stress also works. Still, for those dialects which collapse "schwi" and schwa, schwa is distinct from /ɪ/, so a distinction should be maintained for them. kwami 20:11, 23 October 2007 (UTC)
 * According to Webster's New Collegiate Dictionary, ship in battleship has secondary stress. --teb728 20:44, 23 October 2007 (UTC) (When you say, “It is arguable that English does not distinguish primary from secondary stress,” do you mean secondary stress is not distinct from primary stress or that it is not distinct from unstressed? If the former, I think you have proved yourself wrong.) --teb728 21:01, 23 October 2007 (UTC)
 * Read Ladefoged. "Primary stress" in a dictionary like Webster's is tonic/prosodic stress, not lexical stress. The only thing that makes it different from "secondary stresses" in prior syllables is that it's the last stressed syllable in a word; put the word in context and the distinction disappears. However, "secondary stress" is more generally a conflation of lexical stress and unstressed full vowels: typically any lexical stress before the primary, plus any full vowel after the primary. The latter is what we have with battleship. You could consider stress, full vowel, reduced vowel to be 3 degrees of stress, but L argues that the "secondary stress" in a word like battleship has none of the articulatory characteristics of stress found eg. in the cha of characteristics. For his analysis to work, it's enough to posit plus or minus lexical stress and a phonemic distinction between full and reduced vowels: . kwami 23:19, 23 October 2007 (UTC)
 * OK, though if other linguists disagree then we have to be careful not to take one side too strongly. One thing that troubles me is that I can't think of anything approaching a minimal pair between what under that theory would be /ɪ/ and / ɪ /, and they don't feel to me like different phonemes.  Is it possible that there's actually only one /ɪ/, and a rule for telling whether it's "reduced" or not?  (By the way, I have /ə/ in roses, but do have the weak vowel contrast, e.g. effect with what I'm used to calling /ɪ/ vs. affect with /ə/.)  --JHJ 12:45, 24 October 2007 (UTC)
 * Yeah, I've been wondering that too. The problem for me is that I might have even less of an / ɪ / ~ /ə/ distinction than you do, which makes it hard for me to evaluate these things. (I distinguish Rosa's and roses, but not affect from effect, unless I use spelling pronunciations /æ/ and /i/.) But a word like battleship - I have no problem with giving the last syllable secondary stress, as that keeps things nice and distinct, but if you accept L then you certainly need to conclude that it's a full unstressed /ɪ/. A reduction rule might be possible if it took into account whether the word was a compound or not. But a minimal pair ... The fact that the OED has decided to buck UK tradition on this point suggests that they think there is least a practical reason for doing this. kwami 16:58, 24 October 2007 (UTC)
 * Wait, here's one obvious reason: If some people conflate / ɪ / with /ɪ/, and others conflate it with /ə/, then you're going to mislead half the population if you use either of the latter transcriptions, while / ɪ / is unambiguous. kwami 17:02, 24 October 2007 (UTC)
 * Have you considered using /ɘ/ for reduced /ɪ/? That seems a natural thing to do. Reduction goes toward /ə/ in the chart, but in this case does not quite reach it. &minus;Woodstone 18:40, 24 October 2007 (UTC)


 * That's a good argument phonetically, but graphically it has some drawbacks. In British tradition is used, and some editors object to changing that. In the American tradition ɨ is used.  is in effect a compromise character, graphically similar to both, so it's both intuitive with existing tradition and shouldn't be too objectionable to either. It's now used by the OED, which means we're in good company. ɘ, on the other hand, is so close graphically to ə that I'm afraid people will get them mixed up. kwami 19:53, 24 October 2007 (UTC)


 * But is not an IPA symbol. That's a big step to take. Then I would prefer to stick to /ɪ/. The symbol /ɨ/ is a fairly strong sound on the edge of the diagram. The sound in Russian <ТЫ> does't resemble at all the e in . &minus;Woodstone 20:09, 24 October 2007 (UTC)


 * But ɪ obscures a phonemic distinction, which means we're back to ɨ, which RP speakers generally don't like. (Whether or not Russian /ɨ/ is similar to English /ɨ/ is no more relevant than whether Russian /ʂ/ is similar to Mandarin /ʂ/, but I agree with you about not caring for it.) It's not at all unusual for systems to use a symbol or two outside the IPA. This is really common for affricates, such as λ-bar for . The OED uses, so it really isn't much of a step. kwami 22:25, 24 October 2007 (UTC)
 * No phonemic distinction is obscured. Stress (and syllable structure) is an environment, and distinctions based solely on environment are non-phonemic.  That's an elementary point I teach my linguistics students early on.  It shouldn't be hard to grasp. RandomCritic (talk) 12:14, 6 April 2008 (UTC)

We should stick to IPA, as per the MOS. --Kjoonlee 19:48, 28 October 2007 (UTC)

I'd think that /ɘ/ would be easily confused with the schwa since it is very rare, especially when it appears in isolation. —Michael Z. 2007-10-25 02:15 Z 
 * I've come across a couple unstressed [ɪ]s that are not [ ɪ ]. Here's another: the 2nd y in wysiwyg. kwami 23:29, 7 November 2007 (UTC)
 * Ah, I found one: chauvinism. kwami 01:52, 9 November 2007 (UTC)
 * These are frankly dorky comments. The syllables "wyg" and "ism" are stressed.  English does allow more than one stress per word! RandomCritic (talk) 12:16, 6 April 2008 (UTC)


 * I'm also of the opinion that [ ɪ ], because it's not universal and because it's more of a phonetic peculiarity than a phonemic vowel, that it shouldn't be in the phonemic pronunciation indicators on Wikipedia. That it's also not officially IPA is also important to consider.  Consider also that the syntax involved (and the odd way it turns out looking on the page) means that editors working on pronunciation (those who incorporate this) are bending over backwords to satisfy the misleading notion that one transcription can and should be able to represent both British and American speech.  Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  01:57, 16 November 2007 (UTC)


 * The examples brought up with different versions of unstressed schwa-like sounds are hardly relevant, since we are discussing phonemic transciptions, not phonetic ones. The only relevant examples would be minimal pairs, which I presume are excessively rare. Why can't we desist from using non-IPA symbols and just settle with a choice between /ə/ or /ɪ/ for the few cases where it is phonemically relevant. &minus;Woodstone 06:39, 16 November 2007 (UTC)
 * The OED now makes the distinction, as do most Usonian dictionaries. We could go back to ɨ, but the Brits don't seem to like it. ɪ was a compromise, as well as being current OED usage. Otherwise we get a UK/US split over ɪ vs. ə, which hardly seems important enough to make double transcriptions for. And if people leave off the strike out on ɪ, it doesn't matter much. I think it would be much more useful for us to have multiple transcriptions when we have significantly different pronunciations, such as /ɑːnt/ and /ænt/ for "aunt". But if people want to go with old OED-style ɪ, it wouldn't be too bad. There are probably less than a hundred existing transcriptions that would become ambiguous. kwami 09:47, 16 November 2007 (UTC)


 * We could have it set up so that whether a word is transcribed with <ɪ> or <ə> is based on the same principles as to whether we say color or colour. Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  15:03, 16 November 2007 (UTC)
 * Personally, I think that if a significant number of people make a distinction (though Scots is perhaps too difficult to accommodate), we should make it too, for maximal utility. Spelling differences are completely arbitrary, and writing color for colour doesn't obscure anything. However, if we leave out phonemic distinctions - and for many people the difference in reduced vowels is phonemic - we start defeating the whole purpose of including pronunciation in the first place. If we transcribe a word according to the dialect of whoever starts an article, we lose distinctions, and it would be a little odd to tell people that we have give priority to the dialect of the first author for one distinction, but not for others. Although a GA speaker will generally know from the spelling when to throw in a dropped ar or aitch, an RP speaker will not generally know whether a GA "ɛr" is /ær/, /ɛr/, or /er/ in RP. Then we're back to having multiple dialect-dependent transcriptions for words which don't really have multiple pronunciations. -- kwami (talk) 19:03, 16 November 2007 (UTC)
 * Actually, who cares if ɨ isn't terribly close to its normative IPA value? Neither is ʌ, o, or r. They're all somewhat arbitrary conventions. Shall we just replace ɪ with ɨ? -- kwami (talk) 22:39, 16 November 2007 (UTC)
 * So we use ɨ instead of ɪ ? I'm fine with that. Æµ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi]  03:23, 18 November 2007 (UTC)

My support
I just wanted to tell you guys, especially Kwami, that I'm glad this mini-project exists. Whenever I'd see the IPA code for a word in an article, I'd get frustrated because I hadn't a clue how it should possibly be pronounced. At least with this, I can learn it, adapt it, and expand my IPA and linguistic knowledge. Thanks again! 頑張って！ - Cyborg Ninja 16:42, 24 October 2007 (UTC)

/ɔl/?

 * " and  are also commonly seen. Frequently written /ɔ/ before /l/, as in bole /bɔl/."

To me, the only instance in which it makes sense to write /ɔl/ is for the "al" sound in words such as "although" and "already", which is a very different sound. But a number of dictionaries seem to write it as /ɔ:l/, thereby not distinguishing it from the sound in "ball". -- Smjg 13:57, 2 November 2007 (UTC)


 * It depends on dialect. In RP, you have it in ball, in GA you have something similar in bole. Should be clarified. kwami 21:20, 2 November 2007 (UTC)

Proposed merge
I suggest this article be merged with Help:Pronunciation. We don't need both articles in help space. Since this is the English Wikipedia and IPA is the standard here, it is appropriate for Help:Pronunciation to consist mostly of the IPA English pronunciation key, with mention of audio files and a pointer to the full IPA chart for non-English words. The shorter title is more consistent with WP:NAME. --agr 16:30, 7 November 2007 (UTC)
 * Since almost nothing links to this page directly (most is through the IPAEng template), and no actual article links to the other, I don't see a problem moving this to there, and maybe taking something from the intro. By the same logic, we should move Help:IPA pronunciation key to Help:IPA. kwami 17:35, 7 November 2007 (UTC)
 * I agree. Do you want to do it?--agr 18:08, 7 November 2007 (UTC)
 * Sure. It affects so little we could switch back if anyone objects. kwami 18:53, 7 November 2007 (UTC)
 * Here's the original text of Help:pronunciation:
 * Pronunciation on Wikipedia is most often given using the the International Phonetic Alphabet (IPA). The goal is that phonemic interpretations should not differ depending on the reader's regional dialect. For ease of understanding, fairly broad IPA transcriptions are generally used. Some articles also link to an audio file, usually in the Ogg Vorbis format, that contains a pronunciation of the word in question.


 * A good place for English-speakers to start learning IPA is to consult the IPA chart for English and the more detailed International Phonetic Alphabet for English.
 * The only history after the creation of the article was some added links, categories, & marking and unmarking as a stub. kwami 19:17, 7 November 2007 (UTC)

Wheres the funky o?
Pronounciation of beowulf shows an o with a bent bar under it, but this symbol isn't on this page. —Preceding unsigned comment added by 142.165.246.187 (talk) 20:11, 19 November 2007 (UTC)
 * I've provided a link to Old English phonology since that particular transcription is not a modern English one. The "bent bar" is a diacritic that means a vowel is non-syllabic and that some people use for diphthongs.  Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  20:40, 19 November 2007 (UTC)

IPA Pronunciation Irrelevant To Most Visitors to Wikipedia
Just use the customary pronunciation methods provided in the average dictionary. —Preceding unsigned comment added by 68.202.50.178 (talk) 19:16, 22 November 2007 (UTC)
 * Seconded. IPA is a series of meaningless symbols that nobody understands. I was in a large language class, and they did a survey. Only the professor understood IPA symbols. Sorry, but it's worthless crap. Use the same pronunciation guides that dictionaries use. 64.203.62.162 (talk) 06:33, 19 January 2008 (UTC)
 * There is no such thing, only a gaggle of conflicting formats. You're welcome to add something like that as a secondary guide. kwami (talk) 05:41, 6 January 2008 (UTC)
 * My dictionary uses IPA. --86.148.57.131 (talk) 01:13, 7 January 2008 (UTC)
 * Mine doesn't. Or let's put it another way. What percentage of visitors to wikipedia would be able to look at "ʃaːdənˌfʁɔʏ̯də" (a word I actually need help pronouncing) and be able to pronounce it? 1% 0.01% Compare with the percentage of people that would understand a standard dictionary pronunciation guide, which would be probably between 50%-90%. IPA is as worthless as Esperanto, and it's brain-dead that wikipedia would enforce it. 64.203.62.162 (talk) 06:40, 19 January 2008 (UTC)
 * This is an international encyclopedia, or at least it tries to be. By your argument, if you and your friends don't understand the metric system, or taxonomic nomenclature, we shouldn't use them either. kwami (talk) 07:25, 19 January 2008 (UTC)
 * By that reasoning, we should eliminate English entirely from Wikipedia and standardize on Esperanto. It IS an international encyclopedia, after all. Who cares if less than 1% of the visitors can read it? It's better than the English helping nobody! 64.203.62.162 (talk) 10:34, 26 January 2008 (UTC)
 * You're very close. Check out The Esperanto Wikipedia.  Just as Wikipedia allows an ad-hoc pronunciation next to an IPA one, there exist Wikipedias in numerous languages including English and Esperanto.  As far as I understand, the only person arguing the removal of anything is people against the IPA.  — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  10:57, 26 January 2008 (UTC)
 * While it may be true that seeing "ʃaːdənˌfʁɔʏ̯də" doesn't help you learn how to pronounce it, it also true that no other written form will. So your choices are (1) pronunciation guidance in IPA, which will help some but not all readers, or (2) no pronunciation guidance at all, which will help no one. —Angr If you've written a quality article... 09:32, 19 January 2008 (UTC)
 * It is enlightening that the complaining user does not show how he would indicate the pronunciation of Schadenfreude (yes, I have no trouble reading it). &minus;Woodstone (talk) 14:42, 19 January 2008 (UTC)
 * I did say how it should be spelled -- as we've been using and reading in dictionaries: scha•den•freu•de. Guess what? 90% or more of readers of wikipedia will have no clue how to pronounce the IPA garbage. 90% or more will be able to pronounce it from the dictionary pronunciation. Guess which one is more useful? Make all the mocking comments you want, but IPA is a worthless standard. 64.203.62.162 (talk) 10:31, 26 January 2008 (UTC)
 * Say what?? "scha•den•freu•de" is just the orthography with dots for syllabification added. If you can figure out the pronunciation from looking at "scha•den•freu•de", then you don't need a pronunciation guide at all. —Angr If you've written a quality article... 10:55, 26 January 2008 (UTC)
 * Uh, I would have no idea how to pronounce schadenfreude as it's spelled -- I'd actually guess sha den frood. English isn't like Chinese, the spelling doesn't reveal where the syllables lie. The IPA gobbleygook is less than helpful -- looking at it as a "guide" just makes me pissed off at the article. If it had "scha•den•freu•de", which I guess is what dictionaries use, I would be able to pronounce it. So, well, there you go. IPA is useless, dictionary guides are useful. And I think that if we ran a poll of the common visitors to wikipedia (NOT linguistics majors, who are about the only people that study IPA, at least in America), we'd probably see a 10 to 1 ratio in favor of dictionary pronunciation guides. 71.138.187.65 (talk) 05:17, 30 January 2008 (UTC)

If the people who have the power to decide what is used here feel the IPA pronunciation is appropriate, fine. However, they should acknowledge that most average users will not find it helpful, since they don't understand it. Therefore, for wikipedia to be useful to a larger group of people, alternatives should be supplied. Personally, I think the best tool is a sound recording. If American, British, and Australian pronunciations are different, then provide all three of them. We can all agree that there is no "correct" pronunciation; but surely, typical pronunciations will be helpful to many. Some of this page's discussion sounds like little boy's in a pissing contest. I think the question should be: "Do we want to display our linguistic erudition or do we want to help people in the spirit of an encyclopedia?". I'd vote for the second. 69.92.131.213 (talk) 20:01, 20 January 2008 (UTC)
 * According to a lot of people here, it appears they'd prefer showing off their erudition over actually helping people pronounce a word... Dictionaries have been doing it for years. Ok, sure, British English is marginally different, but even IPA doesn't help with that because it uses guides like "a as in bAll" -- which a Brit would pronounce differently than an American anyway. So then you have to look at the tables which compare British English to American English to Australian English to Scottish English and oh dear I've gone crosseyed. 71.138.187.65 (talk) 05:22, 30 January 2008 (UTC)
 * Inserting several sound files for every word&mdash;while very helpful to the reader&mdash;is a huge undertaking, not likely to happen soon. However there are sound files for (almost) every IPA symbol in the article help:IPA. I concede it will be an effort to reconstruct the word from all those fragments, but it is currently possible to do so for the interested reader. Furthermore, IPA is not that difficult to learn; studying the example tables, a little practice and looking up some familiar words will come a long way in a short time. Finally, for the many that used IPA in their foreign language courses in school, supplying IPA is an immediate clarification of the proposer pronunciation. &minus;Woodstone (talk) 20:22, 20 January 2008 (UTC)
 * Is it really that huge an undertaking? (I'm serious.)  TTS engines have been around for decades now.  The really hard part of a TTS engine is figuring out how a word is pronounced.  IPA does that for us, right?  How hard would it be to use some freely-licensed TTS engine, and batch-create MP3s when someone adds a "pronounced" tag?  I started looking around, but the most popular TTS engines are all marketed as "no more messy IPA", so it's hard to find one that'll boast about it.


 * I came here, like others probably have, because I saw an IPA pronunciation and found it gave me no assistance. Yes, I'm sure the IPA "isn't that hard to learn" if I took the time to do it, and (somehow) kept in practice.  Lots of things aren't that hard to learn.  It wouldn't be that hard for anyone to go find any given piece of information that's in this encyclopedia, but we include it.  On account of it's an encyclopedia.  --JayLevitt (talk) 21:30, 20 January 2008 (UTC)
 * It is Wikipedia policy to allow ad-hoc pronunciation respelling next to IPA transcriptions for the reasons listed above. We shouldn't exclude IPA transcription and we shouldn't (in my opinion) do ad-hoc respelling for foreign words.
 * Also keep in mind that, outside of linguistics articles, knowing the proper pronunciation of things is sort of irrelevant. So you might not know how to pronounce Dostoevsky in Russian or Alia Shawkat; if it's that important to you, learn the IPA.  — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  23:12, 20 January 2008 (UTC)

Yes, let's move toward speech synthesis
JayLevitt brings up a point above that I have thought about too. I have AHD4's electronic version on my PC, and it makes the whole issue of readers needing to understand the written transcription entirely moot by synthesizing audio of every word using speech synthesis (TTS). I'm pretty sure it just reads the AHD-format phonemic transcription and strings together audio for each phoneme into a word. But it works quite well, and it was not expensive for me to buy a copy. Wikipedia should eventually add this feature in addition to the IPA transcriptions (not instead of them). The TTS engine would synthesize based on the IPA as input. (So technically it would not be TTS, but IPATS.:-)) Interested readers could still ponder and squabble over the transcription choices to their hearts' content. But meanwhile the other 99% of WP users (who don't care about linguistics) would sidestep the whole field by simply clicking on the TTS button and listening to the audio. I am not an IT wiz, so unfortunately I can't undertake this feature addition myself. But I hope that eventually someone suitably skilled does so. — ¾-10 23:02, 20 January 2008 (UTC)

I suggest that Wikipedia switch to the pronunciation guide from dictionary.com. —Preceding unsigned comment added by 24.148.244.163 (talk) 01:56, 11 February 2008 (UTC)

Problems 2

 * The key has very severe problems and is misleading and annoying to people (like me) who understand phonetics and phonology. I studied phonetics in graduate school for my master's in Linguistics, using mainly a textbook by Ladefoged. I will point out three severe problems with the key.


 * (1) Phonemes --- abstract entities --- are written enclosed in slashes, like /w/. Phones --- actual sounds --- are written enclosed in square brackets, like [w]. Phonemic representation is for what are called "underlying forms", which can be abbreviated as UFs. A UF has no actual sound, no pronunciation. Phonetic representation is for actual sound, actual pronunciation. Therefore, it is incorrect to misrepresent a UF as a pronunciation, or vice versa. In other words, it is wrong to use slashes for pronunciations. The correct linguistic convention is to use square brackets for actual pronunciations. The key is wrongly presenting phonemic writing as pronunciations.


 * (2) The stated goal relates to providing a "broad" (approximate) representation of the pronunciation (not the UF) by omitting phonetic (not phonemic) detail that can be provided in a relatively "narrow" (relatively exact) representation of the pronunciation (not the UF). Therefore, the IPA symbol for vowel length should not be used, because vowel length is not phonemic in contemporary English. Rather, vowel length is a phonetic detail suitable for a "narrow" representation, but not suitable for a "broad" one. Thus, the key is wrongly including the IPA vowel-length symbol. In actual use in Wikipedia articles, it makes some short vowels wrongly appear to be long.


 * (3) The article does not have even one citation to a source. Mere listing of a reference or two is not adequate. Citations must be made to support everything stated in the article. As it stands now, all statements in the article are subject to challenge and deletion by any editor. —Preceding unsigned comment added by 4.248.13.190 (talk) 19:35, 6 January 2008 (UTC)
 * To (1), the use of slashes isn't entirely consistent in the literature. Slashes are also often used to indicate broad transcription of the surface representation, in addition to their use to indicate unpronounceable phonemes. To (2), whether vowel length is phonemic in English is somewhat controversial, and depends to some extent on dialect. It's clearly not phonemic in General American and Scottish English, but it may well be phonemic in English English (for many people from England, the contrast between shed and shared is one purely of length), and is almost certainly phonemic in Australian English (for whom not only shed/shared but also putt/part and possibly piss/pierce are distinguished only by vowel length). To (3), this is a Help page, not an article, so it isn't held to the same standard of sourcing as an article would be. Here, we're just trying to work out an "in-house style" for transcribing English consistently across Wikipedia articles; we only need enough sources to show that the system we're using has some basis in reality and wasn't just pulled out of Wikipedia editors' collective ass. —Angr If you've written a quality article... 19:50, 6 January 2008 (UTC)


 * The following was written in response to ANgr, before Kwamikagami responded. I got an "edit conflict" message when I tried to save, so I'm trying to insert this in the position where it responds to ANgr.


 * As to (1), are you against using square brackets for phonetic representation? If so, why? Can you cite to pages in an academic publication where slashes are used for phonetic representation? Are you claiming that phonological surface representations are in fact phonetic transcriptions of actual pronunciations? If so, then you need to review your phonology. If not, then the use of slashes for SFs (as well as UFs and intermediate forms) is irrelevant. Why not just use the square brackets? Do you hate square brackets or something? Why are you defending the slashes which you claimed have inconsistent use?


 * As to (2), vowel length is phonemic in Hawaiian, but not in English loanwords from Hawaiian. In the Wikipedia article on Hawaii, the key misrepresents the pronunciations of Hawaii as ending with a long vowel, because the key does not allow [i], but only [iː].


 * As to (3), c'mon man, Wikipedia does not even hold its articles to the "standard of sourcing" that articles are supposed to have. Lack of proper citations is one of the main things that makes Wikipedia an UNreliable source of information. You wrote above that you "only need enough sources". If zero is not enough, then the key has been "just pulled out of Wikipedia editors' collective ass", as you put it.


 * ". . . all progress depends on the unreasonable man." George Bernard Shaw, Man and Superman. —Preceding unsigned comment added by 4.248.13.190 (talk) 21:53, 6 January 2008 (UTC)


 * Anon, (1) is a philosophic issue, not a scientific one. The old 'if a tree falls in a forest' conundrum. Is pronunciation the physical sounds, or their representation in the mind? For foreign pronunciations, we use brackets, unless we also link to the phonology of the language in question, since readers would otherwise be unable to interpret a phonemic transcription. However, since people reading these articles know English, a phonemic transcription is appropriate for English words.
 * Phonemic is not the same as underlying, which for clarity may transcribed between pipes. For example, English cats (given certain theoretical assumptions) would be underlying, phonemic , phonetic . Only the first would be inappropriate as a pronunciation guide.
 * For (2), a phonemic representation may include phonetic detail, depending on how precise the transcriber wishes to be. The point was that such detail should not be included if it does not apply to the phoneme as a whole, such as aspiration. Besides questions as to whether vowel length (or diphthongization) is phonemic in English (which by the way depends on one's theoretical approach as well as dialect), there's a practical consideration: There is ambiguity in a lot of old Wikipedia transcriptions as to whether <> etc. represents or, etc. Therefore, in order to avoid confusion, we decided to not use the plain Latin vowel letters in our transcriptions. kwami (talk) 21:27, 6 January 2008 (UTC)
 * While I agree with Angr and Kwami on the use of slashes etc, I do agree with the anon's concerns about citations in general, which is why I try to be excessive in citing my sources when I include information. While this page itself may not need excessive citations, it doesn't hurt to make many citations in pages like English phonology. — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  22:10, 6 January 2008 (UTC)

Happy tensing
First I want to say I really like this guide and it seems to me like all the right decisions are being made in the end.

I've noticed that there's no distinct representation given for the y at the end of happy or city. I'd like to see /i/ used for this purpose. I think someone mentioned in one of the other discussions that this is the symbol that the OED now uses.

I speak with a New Zealand accent. I'm subject to happy tensing and the weak vowel merger. In my head [i] is an unstressed allophone of /iː/, not an allophone for a terminal /ɪ/. In other words the distinction between /i/ and /ɪ/ seems phonemic to me. That said, I don't believe it's possible to construct a minimal pair to demonstrate whether my [i] actually belongs to /iː/ or /ɪ/. (Because /i/ is always unstressed and always at the end of a word whereas /iː/ is always stressed and /ɪ/ is never at the end of a word.)

By the way, someone gave the i in wysiwyg as an example of an unstressed /ɪ/. I would spell this /wɪziwɪg/.

Ben Arnold (talk) 13:29, 14 January 2008 (UTC)


 * Right now that's in the footnotes. A large number of articles use /i/, and a large number use /ɪ/. Standardizing it to one or the other would require reviewing 20-30 thousand articles (and then having your changes reverted by someone who insists the other way is proper), but leaving it unstandardized isn't likely to cause any confusion. kwami (talk) 18:22, 14 January 2008 (UTC)

/ju/
Per this edit summary, the yod-dropping after is uniform, but I'm not sure how to express that this specifically happens after alveolar consonants without saying, well, "alveolar consonant." Any ideas? — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi] 03:45, 21 February 2008 (UTC)
 * I rewrote the note. Wells's Accents of English (vol. 1, p. 247) gives  as consonants after which later yod-dropping occurs. Spacepotato (talk) 04:25, 21 February 2008 (UTC)
 * All right, that works. — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  06:06, 21 February 2008 (UTC)

Hawaii vs. Hawaiʻi
I changed "Hawaii" to "Hawaiʻi" to more clearly show show the glottal stop issue, and because my understanding is that is the most common way the glottal stop-pronunciation is spelled in he state (and it is the Hawaiian language spelling). Kwamikagami reverted, saying "that d n display well". It displays fine for me. How does it display for others? Libcub (talk) 11:07, 23 February 2008 (UTC)


 * You changed Hawai‘i to Hawaiʻi. The earlier lack of ‘okina was intentional, as that's how most people spell it (and certainly when saying that most people pronounce Hawai‘i as Hawaii.) As an example in English, we might should be an apostrophe instead. For the Hawaiian spelling, coding it (Hawaii) forces it to display differently, but it still doesn't look too good in italics on my browser (FF). kwami (talk) 18:31, 23 February 2008 (UTC)


 * Dictionaries that give pronunciations of "Hawaii" do not report a glottal stop. AHD reports "Hawai'i" as a spelling variant without a distinct pronunciation. Further, the pronunciation of "uh-oh" is not reported to have a glottal stop either (notably, by the RHUD, which uses the IPA). Further, this IPA pronunciation key does not include the glottal stop, and English phonology does not mention it. --Jtir (talk) 19:44, 23 February 2008 (UTC)


 * I and everyone I know, as well as television and the movies, pronounce uh-oh with a glottal stop. The alternative is, which sounds ridiculous. The Hawaii example isn't a good one, but I included it because I couldn't think of anything else. I have heard a number of people on the mainland pronounce it with a glottal stop, but perhaps they'd all visited Hawaii and picked it up there, so it's a bit like pronouncing Paris . Nonetheless, glottal stop is less marginal a phoneme for most speakers than /x/ is. kwami (talk) 02:23, 24 February 2008 (UTC)
 * Yeah, dictionary.com should have put the glottal stop but probably didn't because uh oh is one of those utterances that sort of defies normal English phonology like uh huh (which requires nasality)... though they seem to be indicating the nasality of uh huh in that word's listing. — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  03:00, 24 February 2008 (UTC)
 * dictionary.com is simply copying from other dictionaries, including the Random House Unabridged Dictionary and the American Heritage Dictionary.
 * Merriam-Webster's Collegiate Dictionary explicitly mentions glottal stops in "uh-oh" and provides an audio clip. So this example is sourced and the MWCD could be linked from the help page. AHD also has an audio clip for "uh-oh". Unfortunately, MWCD's entry for Hawaii does not include an alternative pronunciation. However, the audio clip for Hawaii at AHD is distinctly different from the one at MWCD.
 * Hawaii has an informative paragraph on the distinct pronunciations. IMO, "Hawaiʻi" would be a fine example,
 * if the Hawaiian spelling were used to indicate that the intended pronuncation is Hawaiian, rather than English,
 * if the full IPA transcription were given, so readers can see where the glottal stop is used, and
 * if an audo clip were provided that has both pronunciations, so the contrast may be heard.
 * "Uh-oh" would benefit from the same comprehensive treatment.
 * --Jtir (talk) 12:36, 24 February 2008 (UTC)

pronunciation of "ugh"
Re marginal consonants: The pronunciation of "ugh" is sometimes or. RHUD MWCD (The second audio clip at MWCD is especially distinctive.) --Jtir (talk) 19:00, 24 February 2008 (UTC)
 * Another one is "yech". RHUD AHD MWCD --Jtir (talk) 19:31, 24 February 2008 (UTC)

/r/ vs /ɹ/
If this is supposed to be IPA, then why is the symbol for the trilled alveolar consonant used instead of the /ɹ/ symbol that is used for the alveolar approximant which is the sound in very and run? 128.138.64.130 (talk) 22:49, 25 February 2008 (UTC)EF
 * It's phonemic, not phonetic. Read the note. kwami (talk) 23:07, 25 February 2008 (UTC)

vs. )
I'd kept these as alternates mainly because causes fewer complaints when used for place names in non-rhotic dialect areas, or for names when that person doesn't pronounce the ar. A large number of articles use this format, so I think we need to keep these in the chart. Also, the examples that were just removed illustrate the differences between, which may be helpful for some people. kwami (talk) 20:49, 28 February 2008 (UTC)


 * The intention was to establish a simple, uniform guideline for a broad transcription. Of the many combinations with rhotacized vowel, only two had an alternative. This makes for a somewhat redundant and inconsistent system. Examples explaining more than one sound at a time do not belong in a single line in the table, but could be added in explanatory text elsewhere. The utmost simplicity makes for less discussion. It would be possible to add a separate section giving a few alternatives sometimes met, but the guideline should be unambiguous. &minus;Woodstone (talk) 21:12, 28 February 2008 (UTC)


 * Yeah, you're right. I'll put it in a footnote with the other comments. kwami (talk) 21:24, 28 February 2008 (UTC)


 * I've placed /ɝː/ and /ɚ/ back in the key, as we have a large number of articles using these transcriptions. Spacepotato (talk) 23:06, 28 February 2008 (UTC)


 * The key has two purposes. First to help readers interpret the IPA as applied to English. Second, to provide a standardized way within wikipedia how to apply IPA for English. For the second purpose it is confusing to show two alternatives. Editors will wonder which one to use in which case, while really at the intended phonemic level there is no difference. Therefore it is better to keep the key clean and unambiguous, adding a footnote to indicate alternatives that can be found in some articles, that are not (yet) consistent with the guideline. &minus;Woodstone (talk) 23:19, 28 February 2008 (UTC)
 * As the primary purpose of the key is to allow readers to interpret the pronunciations found in articles, it's preferable to show both transcriptions. If it's desired to standardize on one of (e.g.) /ɚ/ and /ər/, the way to do this is to start fixing the articles linking to this page first, and change this page once the deprecated alternative has been purged. Spacepotato (talk) 23:26, 28 February 2008 (UTC)
 * As long as the alternative is in the key, editors will keep adding that form to articles. Removing it now would diminish that effect and allow gradual standardization to a consistent representation of rhotic vowels. &minus;Woodstone (talk) 08:28, 29 February 2008 (UTC)
 * The primary forms have the virtue of consistency with the other rhotic vowels. The alt forms have the virtue of reducing the risk of edit wars. kwami (talk) 00:54, 29 February 2008 (UTC)

Controversial, OR
Kindly do not remove tags because you disagree with them. It will not do to say "this scheme is not controversial". Self-evidently it is. Indeed, it is without support of any kind.

In his ceaseless efforts to promote this pronunciation scheme, User:Kwamikagami writes: "this is how we [sic] indicate English pronunciation on wikipedia".

This is the central problem with this page. Besides the obvious fact that Kwamikagami's "we" doesn't include those Wikipedians who actually know something about linguistics, there cannot be a Wikipedia-specific pronunciation scheme. Wikipedia does not exist to promote some individual's, or even some small group's, idea of what is appropriate symbolism for pronunciation -- whether devised afresh or cobbled together out of several different dictionary schemes. You cannot get around this by saying "This is a help page". It is original research nonetheless and a violation of Wikipedia policy. If it is "only a help page", then on what basis are changes being made to articles all over Wikipedia to make it conform? Putting this on a Help page is apparently an attempt to avoid the kind of scholarly scrutiny to which other Wikipedia articles are, in principle, subject.

I have been asked why I do not propose an alternative treatment instead of just pointing out the errors in the scheme. The answer should have been obvious: there is no single phonemic scheme which will produce underlying representations for all English dialects -- or even just the most commonly spoken ones. That is because the dialects of English are no longer phonemically identical on a synchronic level. You can derive them historically from a common ancestor, but it is quite impossible to produce a "pan-dialectal" scheme which is in the least intelligible to the speakers of any of the dialects. Trying to do so is quixotic, and in the end only does violence to English.

The present pronunciation scheme is a teratological nightmare, which is neither phonemic, phonetic, or pan-dialectal. It makes Wikipedia, or at least certain of its editors look very foolish. If you must have *some* scheme -- which I do not believe is necessary -- then it should be a well-established norm that can be verified by consultation with accessible reference works, consistent with Wikipedia standards. Making something up just for Wikipedia is not acceptable.

RandomCritic (talk) 14:18, 1 March 2008 (UTC)


 * "We" because a number of people worked on this, compared to one person who is philosophically opposed to it. Someone suggested we follow dictionary.com (= Random House). They have  for our <ɔər>, which is inconsistent with the other rhotic vowels (I would prefer they were all parallel with the non-rhotic vowels, but since RP speakers generally don't like that, I think they should consistently use lax vowels); for a high schwa (that is, for both vowels of business);  and  for the long high vowels, which have caused quite a few problems with ambiguity here in Wikipedia; and they leave out the schwa is bottle, button, etc. Other than that, they aren't much different than what we have here. Of course, you could just as easily object to using Random House, since it only gives alternate pronunciations to cover American dialects. However, all of those variants are predictable given the more precise transcription, so we could simply choose that, and then (other than for high schwa) phonemically we'd be back where we started.
 * It is evident from his comments that when RC has in the past said that a pronunciation is 'wrong', he means that it is correct, but that he doesn't like the way it's transcribed.
 * What RC is objecting to is choosing the phonological form in the dictionary that allows you to predict the variants. He prefers to go at it the other way, and indicate only those vowels which are distinct in his speech, subsuming all dialects into his. That violates our commitment to neutrality. (If you insist on promoting your own dialect in your articles, then you'll need to accept that they'll be tagged for violating neutrality, something which you have also rejected.) kwami (talk) 17:55, 1 March 2008 (UTC)
 * You need to check yourself, RandomCritic. Simply because you disagree with people does not mean that they don't "actually know something about linguistics."  Angr, if you haven't realized, is a trained linguist.  I'm not sure about Kwami's credentials and I occasionally butt heads with him but there's no question that he's competent enough to contribute to linguistics articles.  If there's anybody with questionable credentials, it's me and I'm the one who's closest to being on your side.  — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi]  19:04, 1 March 2008 (UTC)