Wikipedia talk:Manual of Style/Pronunciation/Archive 7

Prominence of pronunciation guides in articles
It seems odd to me that the pronunciation of the title of an article, such as a person's name, often seems prominent out of proportion to its significance to the article. As an example, here's the opening sentence of article on the actor Ciaran Hinds: "Ciarán Hinds (pronunciation: /kɪˈɛra:n haɪndz/ or Kee-uh-rawn, with the 'uh' barely spoken; the name is Anglicised as Kieran, pronounced Keer-an where the long 'a' of the Irish is shortened) was born in Belfast." Surely the most important thing is that "Ciaran Hinds is a Belfast-born actor" or some similarly succinct statement of who he is. Why cram the pronunciation of his name, including alternative spellings, etc., into the very first sentence? I'm not suggesting that the pronunciation of names be dropped or suppressed, but it seems odd that the de facto style (i.e., to insert the pronunciation of the name immediately after its first occurrence) is clumsy and awkward. Is this really a part of the standard Wikipedia style? --24.189.116.153 03:19, 15 January 2007 (UTC)
 * Because a pronunciation guide following a word is a convention many guides follow. But for me, the no. 1 reason is that if the term or name is unfamiliar to the reader, then having it up front helps reinforce the proper pronunciation as one reads the article. Only soee articles are going to be as contorted as Hinds. RoyBatty42 17:11, 24 March 2007 (UTC)
 * It is even more awfully prominent when the pronunciation guide is so unusable by the general public that you actually have to put up a freaking "IPA Notice" template right at the top of every article that uses it. Another reason why this guideline should jettison IPA.  (The main reason being that, statistically speaking, nobody understands IPA, and nobody uses it.)  Tempshill 23:57, 12 April 2007 (UTC)
 * Then learn it, peasant. Wikipedia fully covers every IPA character with sound clips and examples from different languages. -- nlitement [talk]  22:20, 21 October 2007 (UTC)
 * "Ciaran Hinds" is just a bad example. Apart from the IPA transcription and the respelled pronunciation, the rest of the information ought to have been put into a footnote. Cheers, Jacklee 22:41, 21 October 2007 (UTC)

when to use?
Sorry if I'm missing something obvious, but perhaps this MoS should provide guidance on when to provide pronunciation help. Certainly obvious names like Bill Clinton don't need it, but where's the line drawn? I notice we tend to include it on non-English names almost exclusively. --W.marsh 21:30, 23 December 2006 (UTC)


 * As a non-English native speaker, I am very happy whenever I find IPA pronunciation, and even happier when I find an audio file. -- Andrea Domenici 131.114.9.221 10:35, 24 September 2007 (UTC)

Use of IPA counterproductive
Consider this a reminder that a world exists beyond wikipedia and the average person who comes to this site does it to find out information based on commonly used tools. And the IPA is a major hindrance.

I'm sure this is probably not a new argument and has probably been debated endlessly before. But I wasn't here then and neither were many who will now use wikipedia. I am a fairly well educated person, I hold a BFA from New York University which I attended on scholarship. So trust me when I say the IPA is not well known as some might assume. When I began to encounter it for the first time on wikipedia, I assumed that it was some sort of Britishism.

Sorry, but what most people who speak English are familiar with is the system used in American and English dictionaries. To use something different even if it is more accurate/flexible/international is akin to deciding that there should be only one international wikipedia written in Esperanto. Why not? It would also encourage people to learn this more useful language rather than English. To belabor the obvious, it wasn't done because people simply want to know what other films Chiwetel Ejiofor has appeared in or what a cudgel is.

It is the same with IPA, which seems to have slapped onto articles in defiance of common sense. At the very least, the use of both the standard pronunciation symbols and the IPA should be encouraged. Considering the concern that editorializing not be allowed in articles, the use of IPA seems like a double standard.RoyBatty42 20:15, 2 January 2007 (UTC)


 * Amen, Brother! I recently had a "layman's pronounciation" (e.g. Shi kaw' go) deleted from an article and replaced with IPA.  On attempting to add back the "layman's pronounciation" (without disturbing the IPA), it was again deleted.  So, I thought I'd chime in here to say that perhaps 95% - 99% of the average Wikipedia visitors have absolutely no clue about IPA and won't take the time to learn it.   It's nice to have a standard, but it certainly shouldn't preclude a common-sense approach to sounding out unusual words.


 * These IPA discussions remind me of an anarchist convention that I attended many years ago. The folks spent hours discussing minor points concerning some theoretical governmental form that would never ever see the light of day and was not in the least practical.  Madman 13:09, 5 January 2007 (UTC)
 * But Madman why should it be exactly your laymans pronunciation of Chicago that was included in the article? Why not mine? I would write it "SheeKAHgow" and my mother would say "sjikago". You cannot make a laymans pronunciation that adheres to any non subjective standard because people read them differently. How many transcription systems for a word should we include then? IPA, yours, mine and my mothers? Or just IPA? I know what I'll choose.Maunus 13:41, 5 January 2007 (UTC)
 * I think that most "layman's pronounciations" (aka pronunciation spellings or respellings) are self-explanatory, and I would not complain about "SheeKAHgow" [except it being incorrect -- : ) -- most Chicagoans say "shuh"].  I would think that any differences in "pronunciation spellings" could be worked out thru the typical back-n-forth Wikipedia process.  Personally, I think a sometimes-crude "pronunciation spelling" alongside the more formal IPA pronounciation would be a service to our visitors.  If it's good enough the for BBC, it's good enough for me.  Madman 18:19, 5 January 2007 (UTC)
 * Sorry, but as a non-English native speaker I find "layman's pronounciations" utterly confusing. I started reading IPA in Middle School, like most Italian pupils, and even if I spent one year in the US, I find it hard to understand a pronunciation from North American dictionaries.  I'm for IPA.  -- Andrea Domenici 131.114.9.221 10:00, 24 September 2007 (UTC)
 * Follow-up: the BBC does not use a pure pronunciation spelling system, but instead uses the typical diacriticals to mark "long" and "short" vowels. It can be found here. Madman 23:40, 5 January 2007 (UTC)


 * Unfortunately there is no set of "standard pronunciation symbols" other than the IPA. If you look at pronunciation respelling for English, you will find that there is quite a bit of variation among the systems used in American dictionaries. Also, most British dictionaries do use IPA. There have been numerous proposals over the years for the adoption of an alternate system on Wikipedia which is more like the systems used in American dictionaries, but all such proposals have suffered from numerous problems:
 * A new system invented for Wikipedia would run into trouble with No original research, and would have the disadvantage of already being known by nobody, unlike IPA, which is standardized and in use around the world by millions of people.
 * Adopting an existing system would be problematic because existing systems are typically designed for one dialect or group of dialects, with no attempt made to encompass pronunciations in dialects elsewhere in the world.
 * Also, using an existing system would cause additional NPOV issues, because existing systems, unlike IPA, make claims about the phonological system of English which may not be widely agreed upon.
 * Unless someone makes a specific proposal which avoids these problems, I don't see what viable alterations there would be to the current Manual of Style recommendations. Nohat 21:11, 2 January 2007 (UTC)


 * I don't want to rehash the issues that Nohat has raised, but I must point out that every one of his assertions has been challenged with what strike me a reasonable arguments.


 * The conventions for applying IPA to English are also vary widely among lexicographers, reflecting both dialect differences and the (to some extent arbitrary) choices involved in constructing a broad, phonemic transcription.
 * A system unique to Wikipedia would simply be a convention, like a hundred conventions in Wikipedia, and not original research, which applies to content. If it did in fact count as original research, so would an attempt to regularize the usage of IPA -- which is currently a dog's breakfast, as has been discussed.
 * The non-IPA systems typically do a much better job of smoothing over dialectal differences than the IPA-based systems. Let me repeat: a much better job. Compare, for example the older and newer systems used in the Concise Oxford. IPA comes into its own, however, in representing dialectal differences - one place where respelling schemes probably shouldn't be used.
 * I'm not sure I understand what claims about phonology might be implicit in a respelling system. Intuitively, I'd guess that respelling systems are less likely to involve phonological claims, since each user is expected to "read" the transcription in his own dialect. But as I've said, I'm not sure I understand the argument here.


 * Just as a sidelight, I'll note that the BBC's internal guide to pronunciation uses a respelling system because, as it notes in the introduction, users can't be expected to know IPA. The users referred to are, of course, professional journalists. I think this pretty clearly refutes the claims, frequently made here, that non-American readers are normally competent in IPA. —The preceding unsigned comment was added by CJGB (talk • contribs) 04:03, 3 January 2007 (UTC).


 * The current standard is, essentially, "there is no specific standard; use IPA symbols to represent broad pronunciations". The problem is that there is no specific proposal for an alternative which is better than the current way we use IPA. Whenever discussions have moved toward picking a conventionalized system—whether based on IPA or one of the respelling systems—discussions always break down when it comes down to the phonological nitty gritty of which phonemes to encode and how to encode them.
 * I think everyone here agrees that it would be nice to have a more specific standardized system to represent pronunciation—IPA-based or otherwise—but discussions in that area have not been very fruitful. All reasonable and specific proposals will be considered. Nohat 06:26, 3 January 2007 (UTC)


 * P.S. the phonological claims I was writing about are like if a system uses the same symbol for the first vowel in father and bother, or the same symbol for the vowel in fun and the final vowel in sofa, or the same symbol for the final vowel of roses and Rosa's. Nohat 06:32, 3 January 2007 (UTC)


 * Perhaps we can auto-translate the IPA into other simpler methods for those who can't easily read IPA (which would be a lot of people, remember there's readers of all ages, all levels of education and disability). I've tried by hand and I'm hopeless at converting IPA into anything useful SHayter 00:32, 15 March 2007 (UTC)


 * I agree that IPA is consistently annoying to me as an educated Wikipedia user, and I think that it ridiculous to expect the normal user to know/care about IPA. I use "normal" instead "average" user because many educated and intelligent people also cannot be bothered to study and learn the IPA. Aristoi 21:00, 22 May 2007 (UTC)


 * The use of IPA is laughable, unless standard English dictionary pronunciation guides are also used. Emperor IPA has no clothes. Lou Sander 15:07, 25 May 2007 (UTC)


 * But there is no standard English dictionary pronunciation guide. — The Storm Surfer 21:56, 25 June 2007 (UTC)

Proposal
Hi. The discussion above (i.e., "Doesn't seem to make a whole lot of sense") kind of faded away. However, I dont see an agreement. So, to continue the discussion, I will start again here.

The project page currently reads: "[other non-IPA pronunciation systems] can however be used in addition to the IPA version so that it's easy for people who don't know the IPA to understand them."

Some suggestions:


 * I agree with the project page that IPA pronunciations should be generally used throughout Wikipedia and I agree that in some cases alternate pronunciation system may be used in addition to the IPA. I suggest that the project page remain as in with respect to this.


 * If the alternate pronunciation systems are allowed in articles to appease readers that do not want to learn IPA, then I think they should be standardized so that everyone will use the same system. Copyrighted systems need not be a concern (as mentioned above) since it is relatively easy to devise a system based upon a few popular dictionary systems. I have played around with something based on American Heritage, Random House, and Merriam-Webster's, which I will paste below. (By the way, although Wiktionary calls their system AHD, it is not exactly the same thing (http://en.wiktionary.org/wiki/Wiktionary_talk:English_pronunciation_key).) Of course, you could also simply follow Wiktionary's scheme. Ultimately, the choice of which symbols is not important. What is important is standardization so that a reader should only have to use one system.


 * I further suggest that the alternate pronunciation guide only be used for general standardized English pronunciations. No English currently has all the historical contrasts and devising something that does so is too much work and generally not encountered in dictionary pronunciations—so we dont need to do this, just follow the dictionaries. Since I get the impression that the anti-IPA-ers are mostly American readers (perhaps a mistaken impression), it may be desirable for the alternate pronuncation guide to be used only for General American English pronunciations. It will probably not be the best idea to use the alternate system in articles discussing dialects and non-English languages (including also the [x] of Scottish, etc.).


 * This new standardized alternate pronunciation system should be tied to the IPA. That is, on its explanation page, it can be explicitly stated that a given symbol is equivalent to a given IPA symbol (i.e. "Let non-IPA  = IPA [æ]"). Doing this will resolve any ambiguity where a certain spelling symbolization may be interpreted as having different sound values depending on the reader's English. Thus, if the word marry is transcribed as [marē] since we would already stated that  = IPA [æ], then it will be clear what the pronunciation is and that we are considering Englishes in which marry does not rhyme with merry.


 * Another suggestion: in order to avoid cluttering up the IPA pronunciation space, the non-IPA pronunciation can be "hidden" with the title code, where if a reader hovers their mouse cursor over the IPA pronuciation the non-IPA will appear. You can use: [IPA] . For example, Chicago (IPA: [ʃɪˈkɑːgo] ). I dont know if this is a good idea.

The proposed pronunciation system is below. For all sounds not listed (such as l, m, etc.), IPA should be used since it is the same as most dictionaries in these cases.

About symbol choice:


 * followed Consice Oxford English in using dh for [ð]
 * vowel symbols are mostly American Heritage except (1) breves were omitted as in Random House & Webster's, (2) Webster's u was used for [ʊ] instead of American Heritage's o͝o
 * since [ʌ] can be predicted as occurring in stressed syllables only and [ə] as only in unstressed syllables, Webster's was followed in using only ə for both [ʌ] and [ə].
 * [ɝ, ɚ] are often analyzed as surface realizations of [ʌr, ər] in dictionaries and phonological descriptions of English. Thus, they are both represented as [ər] as in Webster's.
 * Webster's ŋ was used over American Heritage's ng''
 * since [ɒ] has mostly merɡed with [ɑ] (and [ɔ]) in American Englishes, only a single symbol ä is used. This is the position of Webster's but not of American Heritage and Random House (who have both ä = [ɑ] and o = [ɒ]).
 * accented syllable are indicated with a  '  before the syllable as in IPA and Webster's Online practice. Many dictionaries place the mark after the stressed syllable. Changing this is trivial.

– ishwar  (speak)  22:51, 3 January 2007 (UTC)


 * I think this is mostly reasonable, although I have a few comments:
 * I don't think ô is any better than ɔ.
 * o͞o looks like o?o in Firefox. There is a lot of history of the breve over two characters not displaying well, even in browsers/fonts that handle most of the other stuff fine. How about ū for /u/ and yū for /ju/ (boon [būn] goose [gūs] prune [pr(y)ūn] cue [kyū] few [fyū] union ['yūnyən])?
 * If you maintain the [ɒ] / [ɑ] distinction (using [ä] and [o], at no additional cost), these pronunciations will be much more usable for non-Americans.
 * Nohat 23:26, 3 January 2007 (UTC)

This entire debate regarding making a proprietary scheme in lieu of IPA is missing a fundamental point. English speakers are used to simply seeing pronunciation assistance done in an ad hoc manner. As in, “The Johns Hopkins School of Advanced International Studies, better known as SAIS (pronounced akin to cise in precise)…” Even adding in the most basic markings beyond standard English simply is a nuisance and of no use to 99% of readers. Take for example even the relatively common umlaut vis-à-vis the IPA lexicon; how many English speakers have any clue what that is supposed to do phonetically? The big objection that there are many ways to pronounce Chicago for example can simply be overcome by consensus, and perhaps adding a regional notation if there is a strong local difference. For Example, “New Jersey, pronounced New Joy-see per the state’s accent…”


 * Hi.
 * Yes, maintaining the [ɒ] / [ɑ] distinction is fine and very common and allows for non-American comparisons. A minor issue.
 * I have debated over the oo vs. u because of the different spelling patterns ( . oo (used by American Heritage) seems more common than u (used by Webster's). I used a split system. But, if there is a technical issue, then definitely u is preferred. So, u = [ʊ], ū = [u]. Good.
 * I dont think that ô is better either. In fact, I think it's worse and originally I contemplated using ɔ or aw. There would be 3 different o symbols (o, ō, ô and also oi, or). One issue is that if o = [ɒ], then oi, or should be changed so that these are not confused with [ɒɪ, ɒr]. Using ɔ would fix that problem easily. I used ô simply because it was commonly used and to avoid objection from the anti-IPA faction. If this is not an objectionable symbol to many, then I agree that it would be preferable.
 * Is there any thing else that is "unreasonable"? – ishwar  (speak)  23:47, 3 January 2007 (UTC)

I don't really think a GenAm-only system is appropriate, and having a transcription which only covers one dialect maintains most of the disadvantages of IPA-based systems. Some comments on the system as proposed from a British perspective: --JHJ 09:26, 4 January 2007 (UTC)
 * I don't like the merger of schwa and the STRUT vowel. No British dictionary does this (as far as I know), and it feels very counterintuitive to me.  The OED in fact has a minimal pair: muzz, one pronunciation of Ms.
 * The vowels before /r/ need more explanation. Are we distinguishing merry/Mary ( and ) and serious/Sirius ( and )?  The symbols chosen for pair/care and pier/peer suggest not, which I don't find acceptable.  These distinctions are made in eastern US accents too, and it wouldn't be much effort to add them.  (Treat Mary/care like FACE before /r/ and serious/pier like FLEECE before /r/.  This is what e.g. Chambers does.).
 * Most British accents group CLOTH with LOT, not THOUGHT.
 * I disagree with your transcription of union, which for me has three syllables ( (unlike onion, which has /nj/), but this is probably a dialect variation issue.


 * Hi. If there should be an alternate system that attempts to cover both British and American Englishes, then this should be discussed. (Has it been discussed?) The question is do non-American readers need an alternate to the IPA? I dont see where this has been agreed upon. If no, then there is no need to adjust an alternate scheme to match an IPA transcription of British English.


 * If the answer is yes, then I see that there are two possible ways of doing this. Starting with two IPA pronunciations for the two Englishes, one may simply assign a single non-IPA symbol to an IPA symbol's sound value. For instance, we stipulate: "let k = [k], e = [ɛ], ə = [ə], r = [r]. Then for care we have (Brit) [kɛə] = keə and (US) [kɛr] = ker. Another possible way, which I would consider less preferable, is in certain cases to assign different values for different dialects, e.g. "let X = [Z] iff Brit, = [Y] iff US".


 * Also, I point out that this proposal does not attempt to do anything that the IPA does not do. It only proposes to transliterate IPA symbols into symbols more commonly used in American dictionaries. Whatever problems exist with IPA transcriptions will remain with this alternate system. The proposal also says nothing about how many dialects are to be indicated in given articles (which would also be using IPA in addition to the alternate). Some may want only two (Brit, US), others may want three (Brit, US, Austr). Since this is a separate issue, I have nothing to say about this here.


 * Thus, whether a Gen-Am-only system is appropriate will be determined by first answering if non-US readers need an alternate pronunciation guide. Then, the system would need further adjustments (and I would need help and/or names of British dictionaries that dont use IPA since I am not aware of what is popular in the UK).


 * The [ʌ, ə] convention is that of Merriam-Webster's. However, many dictionaries do not follow this and have separate symbols for [ʌ, ə]. Changing this is trivial. (Regarding variation of [ɪ~ə~ ɪ ] in unstressed positions, this can be ignored as the full range of variation cannot practically be indicated. American Heritage, Random House, Merriam-Webster's all use i (i.e. [ɪ]) in things like Mrs., Ms., -ed, -es. OED lists [ɪ] in the first pronuncation (assuming that this is significant). Kenyon&Knot mention variation in their introductory remarks and list entries with [ɪ] first and [ə] second (first is more common). Thus, I would suggest only [mɪz] for Ms.. At any rate, using different symbols will resolve this issue.)


 * About r is a good question. Let's break this up a bit.


 * For a quick comparison of Mary/merry/marry, look below. What to do with this? The Webster's strategy is to record a "main" pronunciation (merē) and list variation. The American Heritage & Random House strategy is to record them as unmerged (with the use of three different symbols: a/ă, ā, â. OED only has a single form for US. I would not be opposed to using 3 symbols or 2 symbols or only 1 symbol for US pronunciations.
 * Mary — Merriam-Webster's: [mer-ē ~ ma-rē ~ mā-rē]; American Heritage: [mârē]; Random House: [mârē]; Kenyon&Knott: [mɛrɪ ~ mɛːrɪ ~ merɪ ~ mærɪ]; OED: [mɛːri], (Brit) [mɛri] (US); Cambridge Advanced Learner's: [meəri] (Brit), [meri] (US)
 * merry — Merriam-Webster's: [mer-ē ~ me-rē]; American Heritage: [mĕrē]; Random House: [merē]; Kenyon&Knott: [mɛrɪ]; OED: [mɛri] (Brit), [mɛri] (US) Cambridge Advanced Learner's: [meri]
 * marry — Merriam-Webster's: [mer-ē ~ ma-rē]; American Heritage: [mărē]; Random House: [marē]; Kenyon&Knott: [mærɪ]; OED: [mari] (Brit), [mɛri] (US); Cambridge Advanced Learner's: [mæri] (Brit), [meri] (US)


 * For pair and care these are represented as the same either both ...âr or ...er in US dictionaries. Thus, the US pronunciation should be different from the British pronunciation.
 * serious and Sirius are also the same in Kenyon&Knott, Webster's, and Random House but different in American Heritage (using îr, ĭr). I would suggest following only one vowel for both US pronunciations. Again here there would need to be two pronunciation transcriptions.
 * For peer and pier, American dictionaries and OED have the same vowel (so your English, if these dont rhyme for you, would be ignored here).


 * There is variation within both US and British Englishes for cloth. The US vowel is aligned more commonly with thought. There would simply need to be two transcriptions.


 * The transcription above of union (which is from dictionaries and not of my own creation) is correct. The pronunciation you indicate is British. There would need to be two pronunciations given.


 * – ishwar  (speak)  16:42, 4 January 2007 (UTC)


 * I think you've misunderstood my point about pier/peer etc. I certainly pronounce them the same, and so does everyone else as far as I know.  The point was that the current proposal implicitly identifies the vowel in them with that of KIT.  In dialects with the Sirius/serious distinction, this is wrong: it's Sirius which is like KIT, and pier/peer are like serious.


 * I don't really see any point in just transliterating the IPA. For me the main advantage of non-IPA symbols is that they can be interpreted differently by different speakers.  I can interpret a as [a], a General American speaker can interpret it as [æ] and a Northern Cities Shift speaker can interpret it as  (or whatever it is they say exactly).  (Note the variety in the IPA transcriptions you've quoted above; Oxford and Cambridge use different symbols for several vowels.)


 * You might like to have a look at the system described at Help:Pronunciation respelling key, which does try to cover both British and American accents. You could probably modify it to use similar symbols to those used in American dictionaries (and Chambers).  I don't think there's a particularly satisfactory solution on the way, though.


 * --JHJ 17:21, 4 January 2007 (UTC)


 * Ok. As the proposal stands now, that is only transliterating American IPA transcriptions, there is no distinction between Sirius/serious and no distinction will maintained through different symbols.


 * If the alternate scheme is to be used for RP as well as the US, then the words would need to be transcribed differently as they are in British dictionaries. This can be accompanished without the addition of new symbols: serious (IPA: Brit ['sɪərɪəs], US ['sɪrɪəs] / non-IPA: Brit [siəriəs], US [siriəs]); Sirius (IPA: ['sɪrɪəs] / non-IPA: ['siriəs]); peer/pier (IPA: Brit [pɪə], US [pɪr] / non-IPA: Brit [piə], US [pir]).


 * The point of transliterating the IPA is because many readers/editors have commented how they dont like the IPA, wont use the IPA, and oppose its use in Wikipedia. For a while these readers' needs have not been considered. Currently, on the project page, alternate pronunciation schemes are allowed to occur along with IPA transcriptions; however, there is no accepted standardized alternate system mentioned on the project page. Some have objected to a pronunciation scheme based more closely to English orthography that tries to accomodate many different English varieties. Therefore, I have proposed this scheme that does not attempt cover so many varieties, is a little more divorced from the orthography, and is very similar to the several schemes already in use in current popular American dictionaries. It would be better if readers would learn to overcome their reservations about learning IPA symbols for English sounds, but it is not clear that this will be happening.


 * After looking a little at this, I think that the alternate scheme can accomodate RP by the addition of only one more symbol to represent RP [ɜː~əː] as in nurse (which would of course correspond to GenAm [ʌr]). —The preceding unsigned comment was added by Ish ishwar (talk • contribs) 19:49, 4 January 2007 (UTC).


 * To comment further on the mergers above, if GenAm and RP are both transliterated, then there is no need to consider merry/marry/Mary distinct in GenAm pronunciation as this is not GenAm (that is, I would suggest following OED, Webster's, Cambridge). Kenyon & Knott do have marry as only [mærɪ] though, so it may be ok to leave this distinction even though it is not GenAm. Previously, in discussion with Nohat, I added o to accomodate RP & eastern US. However, if RP is also indicated in articles, then I suggest that only ä be used for US since GenAm doesnt have [ɒ]. Readers who do not speak GenAm may be able to deduce their pronunciation by comparing the RP pronunciations, but ultimately these speakers (including me) would be explicitly ignored. – ishwar  (speak)  21:34, 4 January 2007 (UTC)


 * But the American dictionary schemes (well, at least the AHD one mentioned below) and Chambers aren't transliterations of the IPA, and they don't need two transcriptions for words like care with predictable differences between accents. In the AHD scheme, â can be taken to mean  in GenAm,  (or a diphthong) in RP,  in Scottish English, etc., all in one transcription, and a note can make it clear that r isn't pronounced in non-rhotic accents except before a vowel.  I've only had a quick look, but I get the impression that the AHD scheme, with appropriate interpretations of the symbols, is actually better than an RP-based IPA transcription at describing my British English.


 * What I'd actually like is something like the AHD scheme with a table showing IPA interpretations in a selection of accents. Of course you'd still need two transcriptions for less predictable cases like cloth, union and pasta.


 * --JHJ 20:28, 5 January 2007 (UTC)


 * I see. Your suggestion would turn this back toward a system mostly the same as Help:Pronunciation respelling key (HPRK). The difference between AHD and Random House (and probably others I havent looked at, AHD & RHD are identical) and that proposal is that HPRK shows the differences in fern/fir/fur while AHD-RHD do not and AHD-RHD show the differences in hoarse/horse while HPRK does not. Additionally, HPRK has a symbol orr for the first V in orange while AHD-RHD list two different pronunciations. Cases like cloth, pasta are also listed with two pronunciations (union is only listed with one).


 * If you like a type of system, such as HPRK, why was it never implemented as an additional alternate to occur alongside the IPA transcriptions? These systems obvious have a precedent in dictionaries and also in the sister project Wiktionary. —The preceding unsigned comment was added by Ish ishwar (talk • contribs) 00:12, 6 January 2007 (UTC).

IPA vs. American Heritage
Bearing in mind that any system is going to have pluses and minuses, what is the argument for using IPA vs. some dictionary system that uses the normal alphabet--say, American Heritage? Nareek 13:52, 5 January 2007 (UTC)
 * It can be used for all languages, not just English. I doubt the American Heritage dictionary can describe the postalveolar click of the ǃKung language or the mid central rounded vowel  of the Swedish language. --Kjoonlee 07:52, 6 January 2007 (UTC)
 * I think that for many, if not most, non-English speaking readers IPA is much clearer than AHD. IPA is used to teach English (as well as other languages) all over the world. -- Andrea Domenici 131.114.9.221 10:26, 24 September 2007 (UTC)

I agree an old standard should be used. Something native English speakers grew up with, like a dictionary's method - something we were taught in elementary school and can all use without reading any primers. I can't count the number of times I've had to go to m-w.com or my printed dictionary to get a pronunciation of something I was reading about on Wikipedia. For one thing, I can't get IPA to work on my computer, as I apparently don't have the right fonts. For the record, IPA does not display on my stock Windows ME system. For another, I don't want to read the big primer and memorize new symbols. I participated in a national dialect poll a few years ago by a Harvard researcher who asked how you pronounced various words for the sake of determining regional dialects. His poll didn't use IPA or even long or short symbols - he used the capital letter system (whatever that's called (there was an argument about Chicago somewhere on here)) and asked if it rhymed with certain other words. I'm not suggesting that for Wikipedia, but pointing out that here's a Harvard researcher who probably knows IPA and everything else about pronouncing words, yet knows his audience well enough to realize that IPA or other systems will not work for them. I think Wikipedia should learn from that concept.

Here's a simple idea: put a poll on the first page of Wikipedia (http://en.wikipedia.org) and ask people whether they understand IPA. The three answers could be yes, no, and "what is IPA?"...  I think the vast majority of people I know will think you are talking about beer, not a pronunciation system. Obviously, the poll would need to keep track of IP addresses or something to keep the elitists who run Wikipedia from stuffing the ballot box. This whole argument reminds me of "Leet" - a system seemingly designed to be more nerdier than thou... You could rewrite all of Wikipedia in Leet! (please don't respond that there is a Leetapedia or some translator - I don't care) The arguments of IPA being superior do not work. Wikipedia is becoming the standard encyclopedia in the US and standard encyclopedias have pronunciation guides likes those found in dictionaries. There are plenty of examples of superior systems not working for the masses: Betamax was better than VHS; relativity is more accurate than Newtonian physics; the metric system is superior to the Imperial one; Linux and Mac are "better" than Windows - but you have to work with your standard audience...  Formula One cars are "superior" to standard passenger cars, but 99% of us don't  own one or know how to drive one, so designing public roads for that type of vehicle would be silly and a waste of time and resources. I'd bet the vast majority of Wikipedia users don't know what IPA is (or at least can't use it). It's the same reason numerical values on Wikipedia are in the base-ten number system instead of hexadecimal, binary or other such system - base-ten is a commonly known system and requires no explanation even though the use of hexadecimal could be deemed a "superior" system as it would require less storage space for multi-digit numbers. As for the suggestion that the American Heritage dictionary's system (for the record, my favorite is Webster's II with a less-than-one-page pronunciation guide) is insufficient for the postalveolar click, I'd say A) I doubt there are any words in that dictionary that require that sound (whatever it is), and B) in that instance, use IPA and that reader can bother learning IPA to figure out how to say that word. This, again, goes with my elitism and Formula One arguments. I could say, a standard passenger car can't go over 150 mph, therefore the Formula One car is "superior", yet 99.999% of us don't need to go that fast. However, if one needs to go 150mph, they can learn to drive that type of car. It's the same with the postaveolar click - most of us don't need help pronouncing it because we don't know any words that use it (nor are we likely to look up one that does). If someone needs to use that sound, they can read the primer on IPA. As far as the regional distinctions, dictionaries have tackled that obstacle for years: offer two different pronunciations! For example, my Webster's II says basil can be pronounced with both a short and long 'a' sound (they use the lines over the 'a' to denote this) and "laboratory" has four pronunciations. One more answer suggestion for my poll: have a forth answer, "My computer cannot display IPA symbols." So, in summary, I think standard dictionary style pronunciation should be the primary pronunciation guide and IPA should be optional. In the instance where a dictionary / elementary school type system won't work, then exclusive IPA use will suffice (and those readers can learn about IPA). And a poll would help define what the audience really knows. Thanks for listening. 208.64.241.229 21:08, 15 March 2007 (UTC)


 * I think you are quite right that a system should be used that most people can understand. But if people feel that IPA is particularly tricky, that is not because it really is difficult, but just because it has not been sufficiently explained on Wikipedia. (For example, a lot of IPA transcriptions link you to the IPA page, as if what you needed was a dissertation on phonetics rather than a pronunciation guide.) When IPA is pared down to transcribe only English, it uses only a few symbols, most of them like the regular letters used to write English. The ones that are not used to write English are variants of the familiar letters, just as every dictionary has to use some creative mix of diacritics, digraphs, etc. See below on this page for info on a simple chart for IPA and how it might be put into a pop-up on every page.
 * There is no 'standard dictionary style pronunciation". All of them have to be looked up in the end.

--Gheuf 03:13, 16 March 2007 (UTC)


 * But it is hard to learn...  and to get it to display on my computer...  on another discussion board, someone told me it's normal for IPA not to work with Internet Explorer, which, to my memory is something like 80% of the browsers out there.  How can a system that doesn't work on the majority of computers be the best system for Wikipedia?
 * I took your challenge, though, and looked at my Webster's II, which includes a guide for their system as well as IPA and I still disagree with your assertion that it's easy to learn. I think one could learn the basic Japanese "alphabet" (in Roman characters) faster than they could learn the IPA system for English.  Case in point:  the first listing is the short 'a' sound, as in the word "pat".  Webster's chooses to go with the "short-hat" symbol over an 'a' to represent this, just like we learned in elementary school; IPA's symbol is a smashed together "ae".  Reading this, I'm thinking, where have I seen an English word with that symbol in it?  Or at least "ae" written out (not smashed) that is pronounced in a short 'a'?  I can't think of one.  So that's one new symbol to memorize and that I don't think is intuitive.
 * Next sound: long 'a', as in "pay".  Webster's uses an 'a' with a line over it, again, like we learned in elementary school.  IPA uses an 'e'.    Yes, an 'e'.  I try to think of an English word where a single 'e' is pronounced as a long 'a'.  I can't think of one...  That's two things, so far that aren't intuitive and need to be memorized.
 * Third sound: "ar" as in "care".  Webster's uses "ar" with a hat (carrot?) over the 'a'.  Okay, it's foreign enough to me I don't recognize it.  It's not intuitive, as I think it looks like "ar" as in "car", but at least I've seen this symbol before.  Compare that to IPA's symbol:  a sigma (I think, but maybe it's an epsilon) followed by an 'r', or "er".  So there are two representations for this sound, which is odd (and maybe they've narrowed it down since then).  The sigma is definitely something to memorize, and non-intuitive.  The "er" isn't bad, but without a guide I would think it meant "er" as in "jerk".  So, it's no better than the Webster's system for this sound.
 * It keeps going on like that... That was only the first three sounds, I realize, but it just gets worse with more funny symbols including more Greek symbols, integral-signs, funny backwards letters and miniature capital letters.  I can't even guess the name of many of these symbols let alone commit to wanting to memorize them.  Contrast that to a Webster's type system:  the only funny symbol is the schwa, and I remember learning that one in elementary school.  So, by my count, that's one funny symbol for Webster's (the rest are at least in Roman characters), versus no less than 18 weird symbols for IPA's English sounds.
 * Finally, I ask this: if Webster's printed the IPA system next to theirs back in 1988, implying that it existed, yet they chose to go with a dictionary style system as professional printers of a device partially designed to teach pronunciation, why do you think they made that choice?  And why have all the dictionaries I've seen also choose not to use the IPA system?  Why should Wikipedia choose to go against that established grain and try to forge their own, new path?  I think you may have consider that your existing knowledge of this system is influencing your ideas about how difficult this system is to learn.

208.64.241.229 20:17, 17 March 2007 (UTC)

OK, I'm going to throw this out one more time. The great thing about IPA is its generality; it can describe most phonemes from most languages, and dialectal variations within English. The awful thing about it is that it uses weird characters that (1) aren't in a lot of people's fontsets and (2) even if they are, they're hard to remember because most people don't have names for the characters, and a lot of them don't look like letters to which you'd assign such a sound (or indeed, letters at all).

There's a great solution to both problems at once: Kirshenbaum. It's exactly as general as IPA, because it's just a transliteration of IPA into ASCII symbols. But it's much easier (for Anglophones, or in general speakers of Western European languages) to learn and remember, because English phonemes are represented by (sometimes upper, sometimes lowercase) letters that in some contexts are associated with that phoneme.

I know it's not really standard anywhere; that's the biggest argument against it. But it's so easy to learn that it really could be a good solution. (By the way, you won't see how easy it is to learn from the existing Kirshenbaum article, because that article starts with an IPA table and shows the mapping -- what's really needed is a table that shows English phonemes in the context of English words, together with their Kirshenbaum representations. I think such a table can be found at http://www.totally-official.com). --Trovatore 21:08, 17 March 2007 (UTC)


 * The problem I see is that people are replacing other pronunciations that are accessible to the average reader with IPA that is generally not known by the average reader. While IPA may be useful for non-English sounds, most pronunciations can be accurately described by reference to English words and do not need any of the extra, difficult functionality of IPA. —Centrx→talk &bull; 16:25, 24 March 2007 (UTC)
 * I think that IPA is more generally known than North American and British speakers think. I know for sure that IPA has been used for decades in all Italian schools to teach English and other European languages (starting from Middle School, and now from Primary School), and I am pretty sure the same applies to the rest of Europe (excluding UK).  I don't know about South America, Africa, and Asia.  And of course, describing the pronunciation of an English word by referencing another English word may not be really helpful, if you don't know how to read the reference word. -- Andrea Domenici 131.114.9.221 10:26, 24 September 2007 (UTC)

IPA quickhelp templates
I have an idea for making IPA symbols more comprehensible, using tooltips. I have made a template ʒ that contains ʒ, and then a redirect at ʒ as in beige=beɪʒ to the appropriate phonetic page. The discussion of this concept is here: template talk:Ʒ. Without popups, this works wonderfully: someone who doesn't know IPA sees blue text, moves their mouse over the link, then sees the quickie pronunciation help in the tooltip, and if they want to know more, they click and get the appropriate article. The dev version of popups has now been fixed to work with this, but the production version of popups still is not compatible. Again, please do NOT only comment here, instead/also comment at template talk:Ʒ. If and when this starts to get a clearer consensus on whether and how to move forward, I'll be posting this at village pump (policy) and (technical). --Homunq 15:39, 6 January 2007 (UTC)

I think the anti-IPA talk on this page has two bases: dialect issues and comprehensibility issues. This idea does nothing for the former, but goes a long way towards resolving the latter.--Homunq 10:20, 17 January 2007 (UTC)

And a Hover Template for the Guide
Every day, it seems somebody complains about the use of IPA on Wikipedia. Nobody understands it, they say. It is counterintuitive. A respelling system should be used using MAYK-SHIHFT SIGHNZ. Of course, people who know IPA realize the disadvantages of such a makeshift approach, and understand that IPA for English is actually quite simple to use, encouraging those who do not yet know IPA to learn it. But it is the responsibility of those of us who know it to present clearly and accurately the principles on which it is based.

There are many problems facing the new learner of IPA on Wikipedia, but there is one in particular I think responsible for their difficulties, and this is the lack of a real "pronunciation guide" on Wikipedia such as is found in most dictionaries. Most IPA transcriptions, if they link to anything at all, link to the main IPA page which is less than helpful if a mere pronunciation guide is what is wanted.

What we need is a simple chart based on KEYWORDS whose purpose is not to be accurate in cataloguing all the different pronunciations of English (still less listing all the symbols of IPA), but merely to show people how to read the transcriptions. The use of KEYWORDS is important for this purpose, because it allows speakers of different accents to read the transcriptions in their own way. So if I say that "ash" represents "a as in cat", everyone will be able to understand it according to his own accent, and no further phonetic detail will be necessary. The closest thing we have to a pronunciation chart, IPA Chart for English, is accurate as a reference for those studying accents, but unnecessarily detailed for a simple pronunciation key. So I propose something simpler.

1. THE PROPOSAL

I propose a pronunciation chart based on the KEYWORD system that would work across accents. For the vowel "o" as in "goat", it is not necessary to list three different symbols for GenAm, RP, and Australian English detailing the phonetic discrepancies in the realization of this vowel. It is necessary to give only one symbol, /oʊ/, along with the keyword "goat", and every person will read it in his own accent.

Under this system, it is only necessary to list multiple pronunciations of words when they differ in what J. C. Wells calls their "lexical incidence": that is, in layman's terms, when a word has a vowel in one accent that cannot be predicted from its vowel in another.

This is highly preferable to the current system, in which words can theoretically receive multiple transcriptions even if they differ only in phonetic detail, or, in some cases, in nothing more than notational convention. The newer articles in the OED online fall into this trap, so that, for example "poke" and "pack" get two transcriptions each: RP "p@Uk", GenAm "poUk"; RP "pak", GenAm "pæk". This kind of transcription is not only unnecessary, it is misleading, since the presence of alternative transcriptions implies that one transcription would not have been enough. The only purpose of listing pronunciations should be to make clear when these are not predictable across accents. Here again the OED online fails, giving, for example, "was" and "what" as /wQz/ and /hwQt/, with no indication that their vowel in GenAm is the unpredictable /wVz/ and /wVt/.

Let us be even more accurate than the OED and fulfill our goal of making Wikipedia the most thorough source of information in the world.

2. IMPLEMENTATION OF THE PROPOSAL

If this proposal were adopted, each IPA transcription of an English word would be accompanied by the template. The Template does two things: it links to the full pronunciation guide (provisionally, I have linked it to IPA Chart for English); and it causes a pop up to appear listing only the "difficult" symbols (e.g. "dZ" and not "b"). Additionally, the link makes no underline (because of a CSS trick). Here is an example. Ultimately, I would like to have the template link to a simplified chart along the lines I have been discussing. I am working on one at User:Gheuf/Sandbox2: currently it retains the multiple transcriptions used for the vowels, but proposes a "diaphonemic" transcription that would stand for all of them.

3. ON THEORY AND TRICKY ISSUES

Acc. to this transcriptional scheme, each IPA symbol is taken as indicating not a sound, but a correspondence of sounds; not a phone or a phoneme, but what has been called a "diaphoneme." I hasten to add that I do not propose these "diaphonemes" as representation of the speaker-hearer's competence -- I do not suggest that the transcriptions given will necessarily correspond accurately to the mental representations in any one English-speaker's mind. It is rather intended simply as a lexicographical convenience, to allow transcriptions to be readable by many English-speakers from varying backgrounds.

The system works extremely well when there is a simple one-to-one correspondence across accents as in the case of the vowel "o" as in "goat". But we run into problems when accents differ in the phonological systems, that is in the number of phonemes they have and in the positions they allow them to occupy in the word.

As an example of the first kind of difficulty, we may consider short "o". In most American accents (or at least in mine), this vowel has merged entirely with "ah" of "father". So in my accent, "shahs" rhymes exactly with "Boz", "Iran" with "don", "dahlia" (give or take rhoticity) with "jollier". In most English accents, such as RP, these words have different vowels. What to do in a situation like this? I decided to use the UNMERGED variety as the standard: since from it the American pronunciation may be inferred, but not vice versa. This means a certain redundancy for American readeres. They will see "/A as in shah/" beside '/Q as in shod/' : for them there is no difference and the extra symbol is superfluous. But this superfluity may easily be explained in a note, and is necessary for the system to work.

In other cases, however, I have economized on such distinctions. It would, of course, be impossible to indicate every distinction made by every accent without severely bloating the transcriptional system and compromising its purpose. My system does not distinguish between the vowels of "eight" "late" and "wait", nor between "horse" and "hoarse". For new, "difficult" words such as are likely to be transcribed on wikipedia, it is likely that the difference would either not come up or be predictable by the speaker who already makes the distinction. For such a speaker, the symbol /or/ will mean: "pronounce /Q:/ or /o@/ according to context". If such a distinction were needed, the vowel of "origin" could be used for "horse".

As an example of the other kind of difficulty, we may consider the question of vowels before "r", which proved to be so problematic that I listed them in their own special chart. "R" is difficult because it tends to alter the quality of the preceding vowel, and, in a good number of accents (called "non-rhotic") to drop out of coda position altogether. I have opted to retain the "r", not because I believe it to be underlying in non-rhotic accents, but because it is needed for the rhotic ones. A speaker will simply see "ir" as in "here" and associate the signs "ir" with the pronunciation "I@".

I have, furthermore, written the SQUARE, FORCE, CURE, and NEAR vowels as "tense": this is because they contrast in some accents with lax counterparts, as in those accents that contrast "Mary" and "marry", and "spear it" and "spirit" (and even "force" with "north!"). They are of course not necessarily pronounced very tense, although in some accents they are (Scottish English /skwe:r/), and indeed in non-rhotic accents they are almost always transcribed lax -- but they need to be distinguished from the truly lax vowels. Historically, they were all "long" (=tense), since all short vowels before "r" merged in NURSE (with the exception of "start", and, of course, "north" again).

The last kind of problem for this system is words that differ in their "lexical incidence", like "was". Such words (and only such words) will have to be given separate transcriptions for each dialect. This is as it should be, as these relationships are not systematic, although such "anomalous" words often pattern together, whether in small groups (such as "was" and "what"), or larger (such as CLOTH and BATH) -- sometimes only a single word seems to be anomalous (as "pasta", although we might also list "Milan").

4. THAT THIS PROPOSAL IS NOT AGAINST WIKIPEDIA'S GUIDELINES

I would like to show that this proposal is not against Wikipedia's Guidelines, in particular the guideline against original research. The basis for the content of this proposal is of course not original -- it is drawn from standard pronunciation books on English, and J. C. Wells's "Accents of English". All that is new is the proposal of a certain sort of transcription as a Wikipedia convention -- just like all other conventions on Wikipedia, it is not covered by the ban on "originality" in the same sense as contentful articles are. And even if it were covered by the ban, it would still not be original, since it is not my opinion alone that I want to impose, but the consensus of various editors who will, I hope, visit my sandbox page and edit it.

Lastly, I want to calm people who think that this kind of a transcription is "impossible." On this page when similar systems were discussed, some editors suggested that such a transcription had been unreachable even by "professional linguists" and that it was therefore not our place to meddle. But I do not see this as a linguistic-theoretic issue: as I said in the paragraph before, I see this as an issue of a lexicographical convention, just like an other convention on Wikipedia. The great foundation on which it rests, yes, that is the province of linguists -- in particular, I have relied on J.C. Wells's insightful analyses of English accents and his concept of "lexical sets". But the choice of the symbols to represent those "sets" or "diaphonemes" must, of course, be our own. It is not a linguistic, but a lexicographical question. If Wells did not himself assign symbols to his lexical sets by which they might be recorded in international dictionaries, it is not because such an endeavor is impossible or absurd, but because he is not in the business of making dictionaries.

We are.--Gheuf 20:17, 5 February 2007 (UTC)


 * I like your proposal (see my comment here), and I think it's probably better than the one I proposed above. I support this as the standard transcription for English words on Wikipedia. --Lazar Taxon 06:19, 8 February 2007 (UTC)


 * Support. This is the best idea yet that I have seen for the pronunciation problem. It should quell most complaints that IPA is too hard to use if everything you need to know to decipher a pronunciation can fit in that tiny hovering box. I completely agree with all the phonological decisions made here, too. My only suggestion is that in the guide itself, the phoneme symbols don't need enclosing slashes, but they should be joined to the example word with an en dash or similar, . Nohat 08:25, 8 February 2007 (UTC)
 * It's too bad, of course, that the popup guide is broken in Firefox (it truncates it). The only way around it at present is to emulate the tooltip behavior using Javascript and CSS. Nohat 00:50, 9 February 2007 (UTC)
 * Also, if we have to use Javscript, it would be extra neat if we parse the IPA text and only display the key for symbols which actually appear in the transcription. I'll try to hack up a demo of that. Nohat 01:16, 9 February 2007 (UTC)
 * Homunq is working on just that, at Template:IPA_hover. I've discussed with him the idea of combining his template with my pronunciation scheme, an idea he seemed to like.--Gheuf 05:36, 9 February 2007 (UTC)

In case anyone wants to try it out, I have a working prototype of this functionality. Just add to your own User:Username/monobook.js page and then go to any page with IPA pronunciations (in  elements), and try hovering over an IPA pronunciation. It should show an informational popup with just the information for the symbols contained in that pronunciation. Nohat 02:25, 10 February 2007 (UTC)
 * Has any progress been made on this lately? If people seem in broad agreeement, I think a pronunciation guide should be put on the page Wikipedia:Pronunciation, and a pop-up should be made a standard part of the IPA template.--Gheuf 03:15, 16 March 2007 (UTC)

How do we obtain this IPA pronunciation?
Does it need to be sourced? After all, obtaining it from a different pronunciation style given is not exactly something the average reader could reproduce, at least not effectively. -Amarkov moo! 05:49, 15 February 2007 (UTC)
 * The idea behind this is that it would be a convention agreed upon by editors for the use of writing articles in Wikipedia. It is based on other standard IPA transcriptions for English from various dictionaries and books. I cannot understand the sentence "Obtaining it from a different pronunciation style given is not exactly something the average reader could reproduce, at least not effectively." I'm not sure in what sense the reader needs to be able to reproduce the "[process of] obtaining [the guide] from a different pronunciation style given" (?). If he needs to know what the guide is, he'll simply look it up at Wikipedia:Pronunciation. There is more about this in my original posting under the heading "THAT THE PROPOSAL IS NOT AGAINST WIKIPEDIA'S GUIDELINES".--Gheuf 14:40, 15 February 2007 (UTC)
 * I don't think I made my concern clear, but it looks like you answered it anyway. Thanks. -Amarkov moo! 02:08, 16 February 2007 (UTC)

To sum up
To sum up the "anti-IPA" position here: no one is going to learn "approximately 107 base symbols and 55 modifiers" as a prerequisite to using Wikipedia. 70.20.136.170 00:08, 19 February 2007 (UTC)
 * But nobody is saying that you have to. The guideline doesn't say "use IPA and nothing else", it just says that IPA should be provided. -Amarkov moo! 00:15, 19 February 2007 (UTC)
 * And even if IPA were the only pronunciation guides offered, it is no more necessarily to learn every last character of IPA in order to transcribe English, than it is necessary to know every last word in English in order to speak it fluently. The analogy is shaky but the point is true: for most purposes, you will only need to know a small subset of the total number of characters available. The additional advantage of IPA is that if you should encounter an unfamiliar character, its meaning is standardized and can be easily checked at the IPA page.--Gheuf 06:21, 19 February 2007 (UTC)
 * Precisely. It also has the further advantage that it actually can represent the pronunciation of foreign words reliably: 'spelling pronunciations' are shaky at best in this respect.  Day-zha voo is quite a good hint at how déja vu is pronounced in English, but it's so much less accurate than  as a representation of French pronunciation.  This is not to say that spelling pronunciations shouldn't be provided in addition: I happen to think they probably should, provided it's clear that they provide only a limited approximation. garik 14:32, 19 February 2007 (UTC)
 * "oo" in this case would also represent the minimal pair /y/ & /u/ in French—if you want to say "beaucoup" (/boku/) but end up saying "beau cu" (/boky/), you run the risk of highly offending people, and generally being misunderstood (consider also the difference between "dessus" (above) and "dessous" (below). The Jade Knight 08:54, 25 February 2007 (UTC)
 * Offending someone, or in fact slightly mispronouncing any word, is a problem for 0.0000000005% of the user base reading the articles and is not sufficient reason for inserting jargon - that's what it is - into the first sentence of our articles. All people want is to know how to pronounce something, roughly.  IPA does not help anyone pronounce anything, because, statistically speaking, nobody understands IPA, and nobody uses it.  Tempshill 23:59, 12 April 2007 (UTC)

Martin Brodeur
Is pronounced "Mar-tan Bro-door" or something similar can someone help the editors of the article insert that or comment here and I can do it? Quadzilla99 12:19, 8 April 2007 (UTC)


 * Done. See article. &minus;Woodstone 17:38, 8 April 2007 (UTC)

"IPA notice" template
The "IPA notice" template should be removed immediately. I can't believe that on articles like Bebo there's a prominent colored text box at the top of the article advertising IPA. This is so offensive. Tempshill 00:01, 13 April 2007 (UTC)
 * ...what? Why is it offensive? -Amarkov moo! 02:47, 13 April 2007 (UTC)
 * It doesn't belong on an encyclopedia article. This is part of a broader problem, with both IPA specifically and with dictionary items cluttering articles and crowding out introductions in general. With IPA, this pronunciation system is being used to replace reader-accessible pronunciations (rather than simply supplanting the reader-accessible ones, they have been replaced), with the justification that every reader should be forced to link to the IPA table and figure it out when the reader understood it quite well with a sound-alike pronunciation or with simpler pronunciation systems commonly used in English dictionaries. With dictionary items in general, etymologies, pronunciations, alternate spellings, etc. often clutter up the introduction to an article, when they do not actually convey what the topic in question is, which is the purpose of an encyclopedia (in contrast to a dictionary). Anyway, this box is unnecessary and shouldn't be used. —Centrx→talk &bull; 04:48, 13 April 2007 (UTC)
 * What's the difference between supplanting and replacing? garik 07:45, 13 April 2007 (UTC)
 * He means “supplementing". Supplementing the IPA with a simple pronunciation like “BEE-bo” would be useful on the particular article referred to because we want to get the corresponding vowels in different English dialects. As a proper noun, the IPA is useful because it should be pronounced approximately as /'biːboʊ/ in other languages; this is not the case for common nouns where soundalike is probably enough.
 * However, supplementing the IPA with sound-alike pronunciation schemes is an incredibly bad idea for non-English languages, or other contexts where the sounds are the important things. Sound-alike schemes must be limited to showing the pronunciation of English words, otherwise the article will be flat-out wrong (think of something like German schön, which I (but not most English speakers) could sound-alike as shern, or the myriad ways to represenent and interpretations of low (a-like) vowels).
 * As for using schemes that are commonly used in English dictionaries ... the IPA is one. In fact, unlike house formats, much the same IPA scheme is used in many dictionaries. I cannot think of a better scheme than the IPA that fulfils the criterion of being a “pronunciation system commonly used in English dictionaries" (obviously it’s impossible for it simpler than itself).
 * —Felix the Cassowary 14:28, 13 April 2007 (UTC)
 * I agree completely. I think sound-alike schemes are a useful supplement when possible, but make a very poor substitute for the IPA. garik 15:18, 13 April 2007 (UTC)
 * I don’t think you do. You probably missed the last sentence of my first paragraph, elaborated upon in my reply in the next section. That, or you do agree completey, but expressed it in such a way to contradict an important part of my opinion. —Felix the Cassowary 13:41, 17 April 2007 (UTC)

IPA is a completely useless system for the VAST majority
There are people here that go on about IPA and how it should be provided on pages for pronunciation. I'm one to totally disagree with that and the VAST majority of people would find it useless and worse still would find it confuses the issue. I've got a BA in English (summa cum laude), a BEng (magna cum laude) an MSc in engineering (High Distinction) and am finishing an MD (top 2% in class) now. This shows I'm smart and that I've spent a LOT of time in university and done a lot of research. I find IPA impossible to use and so unwieldy that I simply refuse to learn it. Come on guys, think about it, the International Phonetic Association sets rigorous examinations for people learning IPA that have fairly low pass rate (a friend contacted the IPA) even amoungst the people it's aimed at, because most people simply cannot come to terms with it. It was developed for professional phonetic linguists and not for the general public. I guess that leads to the question of whether Wikipedia is for the masses of the few intelligencia elite. It's for the masses folks! Personally I will refuse to use it in articles I write or edit (working on a couple) and would fight to undo any edit that added it to my posts on the grounds that it ADDS confusion when an encyclopaedia is supposed to clarify things not muddy them. Also someone stated above that millions of people use IPA, well I'd like the evidence for that because I personally doubt it tops 250,000 in reality. It's very like Espiranto, an interesting idea to be sure, but it's NEVER EVER gonna catch on outside of a SMALL circle of people, and certainly not with the general public (who are most of wikipedias users!) Sorry to tell it how it REALLY is in the real world folks and hope no offense was taken. fr33kman 01:06, 16 April 2007 (UTC)
 * You estimate that no more than 250000 people use IPA. I am not able to challenge your estimate, but I know that IPA has been used in all Italian schools (starting from Middle School, 6th grade) to teach English and other languages, at least since the sixties. It is likely that the same applies to the other European countries, so I guess that the number of people exposed to IPA during their basic schooling should rather exceed 250000.  IPA is not really restricted to the linguistic intelligencija (I am an engineer, btw). Also, note that all bilingual dictionaries (and quite a few monolingual ones, e.g. the Oxford Advanced Learner's Dictionary of Current English) and grammars use IPA. -- Andrea Domenici 131.114.9.221 12:36, 27 September 2007 (UTC)
 * Thanks for your thoughts on IPA. You are right that the system as a whole is very complex. But as used on Wikipedia, IPA is quite simple. It is most often used to indicate an English pronunciation. For this a very small number of symbols is used, in fact (roughly the same number of symbols as you would need for ANY English pronunciation guide, obviously). A list of these, with example words, can be found at http://en.wikipedia.org/wiki/User:Gheuf/Sandbox2.  If you go to User:Nohat/monobook.js and add the code there to your own User:Username/monobook.js page (in your case: User:Fr33kman/monobook.js), you will see a pop-up box appear any time you hover over IPA text: it will list the hard-to-read symbols in that word and then give an example word to show the sound represented by each symbol.
 * Since you have a BA in English, I assume you have thought about the way the English language is structured and written, and so learning IPA for English should really not be any trouble for you. (It is just normal Roman letters, altered, and employed in a systematic way, after all.) Of course, thinking about language can be tough: but then that might be a reason for eliminating any kind of pronunciation guide at all. It certainly is not an argument for replacing IPA with a ree-SPEH-ling SISS-tuhm. In such a siss-tuhm, how would you indicate the vowel in "cook"? for "kook"?
 * I am not sure what examinations you are talking about, or how often people who take them fail. Depending on the number of people who failed, and the reasons for their failure, the number of failures might or might not show that the exam is very difficult. Depending on the nature of the exam, its difficulty might or might not correlate in some reliable way with the difficulty of IPA itself. The supposedly frequent failures of this unspecified exam might or might not be an indication that "most people cannot come to terms" with IPA. I don't know.
 * I encourage you not to give up on using IPA for English. It is not as difficult as you think it is.

--Gheuf 03:53, 17 April 2007 (UTC)


 * For the English words where it is supposedly simple, there is no need for IPA as soundalike works just fine. Also, your script is worthless for the vast majority of readers. People are not required to register accounts and add Javascript just to use an unnecessary pronunciation system whose purpose is served well by soundalike. —Centrx→talk &bull; 03:57, 17 April 2007 (UTC)
 * Soundalike works fine, yes, but it does not give the exact pronunciation, which IPA does. And exact pronunciation is important information, which should be included. -Amarkov moo! 03:59, 17 April 2007 (UTC)
 * To clarify: IPA does not require Javascript. The Javascript thing is something Nohat and I devised to make IPA easier to learn: it is, in fact, a rigorous form of soundalike.
 * Query: Without using IPA, can you tell me how to pronounce "mullah"? (Please don't try "rhymes with fuller." It is clever but we want something international.)--Gheuf 05:45, 17 April 2007 (UTC)


 * Fr33kman, you say you refuse to learn it, so it’s quite obvious that you should have trouble using it! If you set about learning it, at least as a pronunciation guide for English, you might find it’s rather less unweildy than you think.
 * Amarkov, the IPA as used on Wikipedia does not, should not and could not give the exact pronunciation. It would be difficult for it even to give the correct pronunciation for one dialect!
 * Gheuf, “like FULL-a” works. People might interpret it differently, true, but people can interpret /bet/ differently too: Does it spell “bet” or “bate”? Both are valid interpretations according to the Help:IPA for English page. This is significant with a name like “Oresme”, where the pronunciation is unintuitive, the word is foreign, and both are plausible.
 * If a transcription needs to be provided for a particular word, the IPA should be used to the exclusion of all other schemes for foreign languages (soundalike is misleading); that the IPA should be used in conjunction with soundalike for propernames with English pronunciations; and that soundalike is sufficient on its own for English common words.
 * —Felix the Cassowary 13:38, 17 April 2007 (UTC)
 * Felix, the quiz was for Centrx! Well, it is still open, since "FULL-a" is only a rhyming explanation, not a soundalike respelling.


 * If an agreed-upon standard for IPA were employed, a spelling like /bet/ would not be ambiguous: the vowel /e/ would be looked up in the pronunciation guide and that would be that. (By contrast, soundalike cannot have an agreed upon spelling, since that would be contrary to its "ease of use".) I have no idea why the Help:IPA for English page gives /e/ for FACE along with /ei/. /ei/ is standard here. I think the dual transcription is there because Angr pronounces these words with a monophthong.--Gheuf 16:22, 17 April 2007 (UTC)
 * Sorry to post twice, but I want to point out that Felix's rebuttal of Amarkov relies on an equivocation on the word "exact". A soundalike respelling cannot give the "exact pronunciation" in the sense that it cannot reliably guide the reader in a choice of vowels within his language. For example, the word "mullah" could be respelled as "MOO-lah" or "MUL-lah" neither of which reliably indicates the correct pronunciation /'mʊlə/ to the exclusion of the wronɡ pronunciations /'mu:lə/ and /'mʌlə/. A Wikipedia IPA transcription, on the other hand, fails to give the "exact pronunciation" only in the sense that a certain amount of phonetic detail will necessarily not be recorded. This detail is omitted because not necessary to the purpose of the transcription, and would be lost no matter which transcriptional system was employed.--Gheuf 16:31, 17 April 2007 (UTC)


 * Gheuf, I have no idea how I’m meant to’ve concluded the quiz was for Centrx! It’s indented as a reply to Armakov, but/and doesn’t refer to him or anyone. In any case, the soundalike spelling only needs to give you an unambiguous way to understand how it’s pronounced; it does that.
 * As for /e/ and /ɔ/ and other ambiguous IPA symbols, one still needs to learn the symbols, and they chosen ones will often have different intended meanings for many speakers than the IPA ones. For instance, /ʌ/ probably doesn’t mean [ʌ] for anyone, so people who’ve learnt the IPA properly will be no better off than a completely arbitrary system.
 * And no, my rebutal of “exact” was completely different in intent. Both systems will give too much information (e.g. a contrast of /ɑː/ with /ɒ/ or /ɑɹ/ depending on who you are), both will be ambiguous (e.g. mull/mool vs /bed/). Both can however give you the right information, altho not in the same way because they’re not based on the same premises.
 * —Felix the Cassowary 03:10, 19 April 2007 (UTC)

I AGREE that IPA is useless for the vast majority: (1) What are these symbols? (2) Why can't I just see the word phonetically? (3) The computer can't display them properly! (4) The IPA "characters" are squished together making them hard to read unless one enlarges the page first! (5) This is supposed to be English, not some foreign gobbledygook! —Preceding unsigned comment added by 75.179.5.126 (talk • contribs)


 * 1) and 5) Can you suggest a way of producing an accurate pronunciation guide of, say, Khrushchev without using non-English symbols? 2)Eh? This is seeing the word phonetically. 3) and 4)If this is a general problem, then this is worth doing something about; it's orthogonal to the question of using IPA in the first place though.


 * And please sign your posts! garik 19:52, 26 April 2007 (UTC)

IPA is not useful for the vast majority, I think that is clear to anyone who looks. If people *did* learn it then we'd all be happy and pronouncing things correctly. But they will not, and so a sound-alike is required. Telling people to learn IPA or saying it is only a limited amount of symbols they have to learn is beside the point. People won't take the time. I think that Wikipedia ought to be approachable to the masses. IPA is not. I agree that exact pronunciation is important for the few who will know IPA, so I think side-by-side sound-alike is really the best way. Aristoi 21:12, 22 May 2007 (UTC)
 * See my points below. There's nothing wrong with having both. garik 21:59, 22 May 2007 (UTC)
 * Learning to read at all is much harder than learning IPA. Maybe all Wikipedia articles should be books on tape.--Gheuf 16:01, 23 May 2007 (UTC)
 * Hopefully you're just being sarcastic, because it's that kind of elitist attitude that got IPA as the only allowed pronunciation method on Wikipedia to begin with. Aristoi 17:10, 29 May 2007 (UTC)

I don't agree. Don't forget that people from many countries come to Wikipedia. You might learn American Heritage in school, but for other people it is as uselless as IPA. Any writing system will be useless if you REFUSE to learn it. When I learned English, I used IPA, and IPA and IPA-derived alphabets are often used in language learning. And it's not hard to learn, look here, most of the symbols are equal to letters most often used for that sound. 88.101.76.122 19:43, 11 August 2007 (UTC)
 * Yes, certainly if we are to give pronunciation at all systematically (and it seems obvious that we should), the IPA is by far the best choice. garik 23:52, 11 August 2007 (UTC)

Not a fast prescription?
Centrx,

You've added a generally to "Pronunciation in Wikipedia is indicated using the International Phonetic Alphabet (IPA)." noting it not to be a fast prescription ... well it looks as if it had been until the 22nd of March 2007. Jimp 08:57, 17 April 2007 (UTC)


 * The problem is it has been interpreted by some to mean pronunciation in Wikipedia is indicated only by the IPA. Other pronunciation systems should not be removed, and it is perfectly allowed and furthermore appropriate in many or even most situations to use other pronunciation systems. The wording is misleading. —Centrx→talk &bull; 03:14, 25 April 2007 (UTC)


 * I don't see the addition of generally clearing this misunderstanding up in any way. Instead, as I read it, the generally seems to imply that some other system can be used instead of as opposed to in addition to the IPA. This is, of course, what some people want but is certainly not what consensus seems to favour (rightly in my opinion). Jimp 04:11, 25 April 2007 (UTC)

Centrx,

I've reverted your addition of "and other systems" which you added with the comment "see talk". See what ... this discussion? I wouldn't call this discussion evidence of any broad consensus. Your "and other systems" may seem small and inoccent to some but I see it as a major change in policy. Consensus, as I read it, favours the IPA. Whereas "the IPA and other systems", on the other hand, is basically equivalent to "anything goes". Not any goes. Pronunciation is indicated primarily by the IPA. SAMPA was used in the past but is being phased out and the only other system in use is Kwami's. This latter system, however, faces somedeal of opposition. These are the only systems in use of which I'm aware nor am I aware of any consensus for the approval of any other system for use. Policy is not "use any system you like" its "indicate pronunciation primarily by the use of the IPA supplimenting this where appropriate by additional means" where "additional means" means other ways of indicating pronunciation such as "rhymes with ...", "with stress on the ... syllable", "pronounced with the same vowel as in ...", etc. but not, in general, by means of rival phonemic/phonetic transcription systems. Jimp 00:19, 27 April 2007 (UTC)
 * Sound-alike pronunciations are common and appropriate for a general-purpose encyclopedia. Saying "Pronunciation in Wikipedia is indicated using the International Phonetic Alphabet" is constricting; pronunciations are not indicated only using the International Phonetic Alphabet. —Centrx→talk &bull; 00:25, 27 April 2007 (UTC)
 * Sound-alike pronunciations don't constitute a system. "Pronunciation in Wikipedia is indicated using the International Phonetic Alphabet" may be too constricting but "Pronunciation in Wikipedia is indicated using the International Phonetic Alphabet and other systems" is not constricting enough.  The American Heritage Dictionary uses its own system the use of which there is no consensus to accept on Wikipedia.  Do I seem pedantic about wording?  I don't believe I am being so.  I'd say that it is of great importance on a page as (potentially) influencial as this to get it right especially in the opening sentence. Jimp 02:27, 27 April 2007 (UTC)
 * This page is essentially only a guideline for how to use IPA. Maybe rename it to Wikipedia:Manual of Style (IPA)? —Centrx→talk &bull; 04:38, 30 April 2007 (UTC)
 * This page is a guideline for how to use IPA because pronunciation in Wikipedia is indicated using the IPA; not "generally", not "the IPA or some other system", not "can be" but is indicated using the IPA. If you want policy changed, the way to do it would be through discussion rather than just changing policy pages on your own.  I think I see your point but find your rewordings too weekening of the policy.  Perhaps it is your intent to so weeken policy.  The weekening of policy is not a bad thing in & of itself but in this case I think it is.  We have a standard system which works, sure some might have to refer to the chart but this is the same for any system.  All three of your changes would be to invite the proliferation of a plethora of different systems making the encyclopædia harder to use.  These rewordings are real changes in policy and not insignificant ones at that.  Changes of this type should first be discussed.  Have they been?  There's only you and I so far and I've opposed all three of them ... Jimp 09:57, 30 April 2007 (UTC)


 * I have seen many articles where pronunciation is not indicated using IPA, so you need to define what you mean a little more specifically. In other cases, I see that people have replaced good sound-alike pronunciations with IPA, citing this manual of style as though this manual of style prescribed the use of IPA instead of other pronunciations. Replacing a system that every reader understands with a system that few readers understand in this general-purpose encyclopedia is not a working system. So, would you propose some solution to this rather than denying these actually rather minor changes which seem to me to be perfectly reasonable and to follow directly from the problems that you appeared to acknowledge? Also, drop the confrontational attitude. —Centrx→talk &bull; 05:01, 2 May 2007 (UTC)


 * I intend no confrontational attitude, sorry if that's not how I'm coming off. You write that you've "seen many articles where pronunciation is not indicated using IPA," do you include articles where pronunciation is not indicated at all or do you specifically mean articles where it's indicated by other means?  There are plenty of the former & rightly so.  You mention the replacement of "good sound-alike pronunciations with IPA" and go on to mention the replacement of "a system that every reader understands with a system that few readers understand".  Yes, it's time we start defining what we mean.  Whilst it may be true that the IPA is a system that few readers understand and it may also be true that sound-alike pronunciations are, on the other hand, widely understood; I would argue that there is no system that every reader understands.  Though it may seem I contradict myself consider this: do sound-alike pronunciations constitute a system?  Throughout this discussion I've used the word system to refer specifically to phonemic/phonetic alphabets such as the IPA, SAMPA, the schemes used in various dictionaries, etc.  I think that this would be the usual sense of the word in this context.  Such descriptions as "ochre (rhymes with poker)" or "poutine (sounds roughly like poo-tin)" employ no such system.  This is what I've been taking "sound-alike pronunciations" to mean.  The thrust of this policy, as I read it, would be to allow non-systematic descriptive means such as these as supplimentary to the IPA which is still the primary means of indicating pronunciation and the pronunciation system used here.  Yes, I acknowledge the problems you point out but haven't felt that your rewordings were reflecting policy as I read it.  I am considering the wording of my version of the solution to this but how about this idea: put one of those nutshell boxes up the top of the page?  Jimp 16:34, 6 May 2007 (UTC)


 * How about something along the following lines (to replace the first paragraph)?

The system used in Wikipedia for phonetic and phonemic transcriptions is the IPA. Where appropriate, it is also permissible to use descriptions in words or additional transcription systems. For ease of understanding, fairly broad IPA transcriptions are usually used.


 * I think that this captures the gist of the policy sort of okayly. Jimp 05:11, 10 May 2007 (UTC)

Why should IPA be used on Wikipedia when most readers don't know it, whereas sound-alike is much more understandable by more people and is equally effective for most words? —Centrx→talk • 19:01, 17 May 2007 (UTC)
 * Because it's not equally effective: first, sound-alike systems are nowhere near as accurate for non-English words. Second, sound-alike systems can be ambiguous and unclear even for English words. Third, remember that English Wikipedia caters for non-native speakers as well; sound-alike is not necessarily as obvious to them as you think. All the policy says is that the IPA should be the standard – this way we have a single system for English words and foreign words – there's nothing to say you can't use a sound-alike system in addition to give an approximate guide for people who don't have time (or can't be bothered) to check the IPA chart for English. In fact I think we should.  garik 21:12, 17 May 2007 (UTC) edited by garik 21:14, 17 May 2007 (UTC)


 * "Because it's not equally effective": I'd point out that IPA is very ineffective because most users do not know it and will not learn it. It may be more accurate but that doesn't really matter.  A sound-alike system is required to reach users. Aristoi 21:06, 22 May 2007 (UTC)
 * Again, see my points below. Why do people seem to see the two systems as mutually exclusive!!? (OK Aristoi, so you don't, but nor do I, so where do we disagree?) We can have IPA transcriptions for people who want accurate information, and a sound-alike system for people who just want a vague idea. garik 16:45, 23 May 2007 (UTC)
 * Garik, I'm not sure that we are disagreeing.... I'd be fine with them side-by-side (I know you know). The problem I see is the reports that some zealous Wiki editors are removing any non-IPA guides in articles.  I think that the majority of conflict would be resolved if we could all agree to allow for multiple systems, or even just 2 systems (IPA and something) with the rest being converted to these when found out in the wild. Aristoi 17:14, 29 May 2007 (UTC)
 * I agree. I wonder how many editors are really as zealous as they appear, however.  I suspect that some on both sides see the entire argument as being a matter of "which is right?"  But I do think we should stick to two systems (otherwise pronunciations guides just get out of hand), that one should be the IPA, and that the other should be simple enough not to require a key. I explain the kind of thing I mean below. garik 17:19, 29 May 2007 (UTC)

Audio Pronunciation Policy
I have noticed that many words that appear in both Wikipedia and Wiktionary have audio pronunciations in Wiktionary but not in Wikipedia. Is there a policy regarding reusing the audio in Wikipedia, or is it just a case of adding the Audio-IPA template as people see fit? - Chris Wood 20:37, 18 April 2007 (UTC)


 * Wikipedia really needs audio pronunciation files (mp3 or some other common format) for most words. IPA by itself is OK but not sufficient. I had hoped that there would be some kind of program that could generate audio automatically from IPA spellings (it need not sound like "natural" speech-- just string the phonemes together), but so far I haven't found one. This is the sort of thing that computers should be good at!


 * Failing that, maybe some of you IPA wizards could just record some. Again, they need not sound professional, just be good enough to demonstrate how some obscure word should be pronounced. I think most computers have the capability, if you plug a microphone into them. It shouldn't be any harder than editing a wiki page. If there's already an audio file but you don't think it's good enough, make a better one, just like anything else on the page. Rbean 23:58, 12 May 2007 (UTC)

Pronunciation needed
Is there a "pronunciation needed" template for putting after unusual words in articles? This would be meant to encourage regular visitors to contribute audio files if they knew the proper pronunciation. — Omegatron 03:47, 4 May 2007 (UTC)


 * See Template:Pronunciation needed. — Omegatron 23:36, 12 May 2007 (UTC)

Pragmatic approach to pronunciation proposed
I think we need to recognize a couple of basic truths when it comes to pronunciation. One sad truth is that the overwhelming majority of our readers will never bother to learn IPA. But another, happier truth, is that for the vast majority of pronunciations in non-linguistic articles, only one or two sounds actually need to be explained. That is why I propose using something like enphon. I realize this is not fully developed yet, but I think with a bit a work this could help a lot in making pronunciations more understandable to laymen.--Pharos 00:26, 10 May 2007 (UTC)
 * Looks very similar to another idea currently floating about. See
 * 
 * Template talk:IPA hover
 * Jimp 01:46, 10 May 2007 (UTC)


 * Why not just directly include the sound-alike representation rather than going through this rigmarole? —Centrx→talk &bull; 19:00, 17 May 2007 (UTC)


 * Look, if we're talking practical solutions: let's just settle on giving pronunciation in IPA and sound-alike. It seems to me to be the obviously sensible route.  The first is very precise for people who need that, and the second is a quick approximation for the cursory reader.  Let's take the name Mitterand as an example. It might read something like: Mitterand (pronounced, approximately "meeter-on").  The sound-alike pronunciation is clearly not a good guide on how to pronounce the name in French – that's provided by the IPA transcription – but it does at least let the reader know it's not pronounced to rhyme with "bitter hand".  And yes, some sort of template that links to an IPA guide couldn't hurt, and could be a very good thing.  garik 16:24, 19 May 2007 (UTC)


 * I'm not sure whether "rigmarole" refers to using IPA, or to arranging for a standard pronunciation key and a quick way to access it. Obviously, sound-alike would need the latter as well if it was going to be comprehensible.--Gheuf 17:51, 19 May 2007 (UTC)
 * Possibly not, though I took Centrx's point to be paraphrasable as "since using the IPA would involve such a rigmarole, then we should just use a sound-alike system" (I may well have misunderstood him). In any case, if the kind of sound-alike system envisaged also needs a key, then there's little point including it along with the IPA. The kind of sound-alike I envisage is one that would be as instantly understandable as possible for a cursory reader.  So "déja vu" would be represented as "day-ja voo", Gorbachev as "gore-ba-CHOF".  This seems to me to be as unambiguous as such a system can get.  Of course they're only very approximate, but that's the point: if we were to try to devise a system that represented accurately any sound we might need to represent, well... that's the IPA! garik 19:29, 19 May 2007 (UTC)
 * Presumably the purpose of any respelling system is to indicate the pronunciation for people who do not already know it. A soundalike with no key is almost useless unless you already know the pronunciation anyway. For example, from "meeter-on" and "day-ja voo" I would derive the pronunciations /'mi: ɾəɹ ɔ:n/ and /'deɪ dʒə vu:/ if I didn't already know that those words are in English usually pronounced /mi: teəɹ 'ɑ:n/ and /deɪ ʒɑ: 'vu:/.--Gheuf 19:44, 20 May 2007 (UTC)
 * I think you're missing my point. The advantage of having a sound-alike pronunciation in addition to the IPA is that it's better than nothing for people who don't know the IPA and don't have the time to check with a key.  So someone reading "meeter-on" would be in a better position than before inasmuch as they'd now be aware that the name doesn't rhyme with "bitter hand", as they might have imagined  (I would also add that Mitterand is usually pronounced more like /'mi: təɹ ɔ:n/ than /mi: teəɹ 'ɑ:n/ in the UK).  If you want to know how déja vu is properly pronounced, then you really need the IPA transcription; if you haven't got time for that, then "day-ja voo" is at least better than nothing.  I wouldn't expect you to derive the correct pronunciation from "day-ja voo", but you can at least get something closer to that pronunciation than if left to your own devices (/'deɪ dʒə vu:/ is far better, for example, than /dɛdʒə vu:/).  So you get what you pay for, so to speak.  The purpose of the IPA transcription is to indicate the pronunciation for people who do not already know it; the purpose of a sound-alike system is to point people in the approximate direction of the pronunciation if they haven't got time to find out properly. garik 20:11, 20 May 2007 (UTC)
 * I agree with garik on this one. If a sound-alike system required a key and looking it up, we're back to IPA.  I support something that a normal English speaking user would recognize without the need to cross-reference it.  It is not expected to be 100%, or even 95%.  That would be the role of the IPA.  The sound-alike would be a ballpark translation for casual users. Aristoi 17:19, 29 May 2007 (UTC)

Query about MoS text on Pronunciation
A complete copy-edit of the MoS is being gradually conducted. The whole section on pronunciation is only one sentence.

"Pronunciation in Wikipedia is indicated using the International Phonetic Alphabet (IPA). For ease of understanding, fairly broad IPA transcriptions are usually used."

Can someone tell me what "fairly broad" means here. Is there anything else that is important enough to make it into this section (we need to be parsimonious WRT to including more information). Tony 01:01, 9 June 2007 (UTC)
 * "Fairly broad" is a vague expression, probably included to encourage people not to indicate unnecessary phonetic detail in their transcriptions. For example, if you were giving the pronunciation of "Thames" you should, according to the instructions, write /tɛmz/, not /thɛmz/ : the transcription with superscript "h", indicating aspiration, is unnecessarily narrow in that it gives too much phonetic detail for the purpose. Without that requirement, editors might complicate their transcriptions in the interest of "accuracy" -- a process that would, in principle, never end, and through which it would be impossible to reach consensus.


 * (In phoneticians' terminology, the narrower a transcription is, the more phonetic detail it has; the broader it is, the less detail it has. A narrow transcription is useful for the comparison of different regional accents, or for a class on phonetics; a broad transcription is useful for a dictionary entry whose purpose is to show people which phonemes are in which words.)


 * What does "we need to be fairly parsimonious WRT to including more information" mean? That we should have more information, or less? --Gheuf 04:35, 9 June 2007 (UTC)

A while back there seemed to be some consensus support for the idea of using a dictionary standard like Gimson phonemic IPA (see archive 6 of this page), at least among those not opposed to IPA altogether. (More info at Pronunciation respelling for English) But we couldn't really work out what such a standard would be, which would work for British, North American, Australian, Indian, etc. forms of English. But perhaps it would be worth mentioning the idea. —Michael Z. 2007-06-25 23:02 Z 

Non-English
The IPA, in case you've forgotten, is the International Phonetic Alphabet. Wikipedia's scope is not limited to English words, nor English subjects. Consequently, we must use the IPA for all transcriptions, including English words and sounds. --Kjoonlee 21:37, 8 August 2007 (UTC)
 * I agree though, as I've said above, I see no problem with using some simple sound-alike system as a rough and ready guide in addition to an IPA transcription. garik 22:27, 8 August 2007 (UTC)
 * Which will be used for all languages, or just English? And would we choose an existing system, or would we come up with our own? If we choose an existing one (just one), wouldn't that be POV and unfair to all the others? If we come up with our own, wouldn't it be OR? The IPA is the only truly international system, and it's not OR. --Kjoonlee 08:00, 9 August 2007 (UTC)
 * My point, as I've tried to explain above, is that we shouldn't use any system at all. If we did, there would be no advantage to it: you'd have to look up how it works just as you do the IPA.  I envisage something entirely ad hoc.  At best it would only point readers in the right direction, and would not provide anything like the accuracy provided by the IPA.  But that's OK.  That's what we have the IPA for.  And it would be for all languages.  The example I give above is: Mitterand (pronounced IPA: [mitɛˈʀɑ̃], approximately "meeter-on").  You'll notice that "meeter-on" only gives a vague idea of how the name is pronounced.  That's fine.  That's the point.  If people don't care enough to look up the IPA characters, then they presumably don't care very much about getting the pronunciation right.  But, as I say above, it's far better to come away rhyming the name with "meeter on" than with "bitter hand". So I repeat: a sound-alike system would not be as good as the IPA, but that's not a problem.  It's not supposed to be.  Second, it should not involve looking up anything on a table. It should be as useful as it can be on one glance and no more. garik 09:57, 9 August 2007 (UTC)
 * That would only work for English speakers (readers) who speak in the exact same dialect as the writer. For all the others, it would be misleading and incorrect. So I don't think we can use that anywhere at Wikipedia. --Kjoonlee 10:11, 9 August 2007 (UTC)

And if it's ad-hoc, it would be OR. WP:NOR. --Kjoonlee 10:11, 9 August 2007 (UTC)


 * Well, first of all, it's aimed only at English speakers, native or not. If someone's reading English Wikipedia, it's fair to assume they have some knowledge of English. Second, I think the "same dialect" thing is rather overstated.  It is a problem for any pronunciation guide. I fail to see how "meeter-on" would be a less clear guide to people who don't speak Welsh-influenced British English.  A north American is likely to pronounce it with an alveolar flap, instead of the alveolar stop that I tend to use.  Some Londoners might pronounce it with a glottal stop.  So what?  That's probably how they'd pronounce it anyway.  Would they write their sound-alike guides much differently? If they care that it's actually pronounced with an alveolar/dental stop, with a low VOT, compared with English, then they can use the IPA transcription.  I stress: the only purpose of an ad hoc sound-alike approximation is to give cursory readers a vague idea of how the word is usually pronounced.  Obviously there will be times when a rhyme works in some places but not others, but these are often times when we need more than one IPA transcription too.


 * In fact, I think the only difficulty is the one about it being OR. However, I actually don't think it's much more OR than an IPA transcription.  Unless the latter is derived from a source like a dictionary, then it's the work of whoever put it there.  The symbols themselves may be part of an internationally respected system, but their use in writing transcriptions remains impressionistic.  If we apply the OR rule too stringently to pronunciation guides, we'll need to add citations ever time we tell people how to pronounce something.  This is another reason why a sound-alike guide should be ad hoc.  That way, it's barely more OR than using the IPA. I think it's actually more OR if it's not ad hoc! garik 10:35, 9 August 2007 (UTC)


 * I'm not saying that the transcriptions themselves will be OR. I'm saying that the choice and use of an ad-hoc system would be OR and POV. --Kjoonlee 23:13, 9 August 2007 (UTC)


 * I'm saying the transcriptions themselves will be OR if they aren't cited. I'm not sure if this is a problem with pronunciations or a problem with Original Research policy; a transcription from an audio recording, for example, is probably better than no pronunciation at all or a transcription based on however the author happens to pronounce it.  — The Storm Surfer 18:28, 11 August 2007 (UTC)

Garik, if you're not convinced about English dialects being a problem, please have a look at the diphthongs at IPA chart for English. --Kjoonlee 23:22, 9 August 2007 (UTC)
 * There's also the Northern cities vowel shift, Canadian raising, Phonological history of English high back vowels and what not. There are probably dozens of more examples at Wikipedia. Category:Splits and mergers in English phonology. --Kjoonlee 23:27, 9 August 2007 (UTC)
 * You seem to be missing my point. It doesn't matter if two people read the same ad hoc pronunciation guide and come away with different pronunciations. If that doesn't make sense to you, please reread what I've written. It would also be useful if you gave me a specific example of a word that you think would cause problems with an ad hoc guide. garik 16:16, 10 August 2007 (UTC)
 * But it should matter. Examples coming up.. Not to mention non-English problems with "Sju hundra sjuttio sju sjusköterska i Kista" in Swedish, with . 딱지 in Korean, with fortis . English approximations can't describe Korean 딸; ad-hoc it would be tal, but then, that would sound like 탈, a totally different word! --Kjoonlee 07:32, 11 August 2007 (UTC)
 * Korean 셈 and 샘 are two different words, which are pronounced and  respectively. Ad-hoc, they'd both be "sem", and sem is pronounced differently for each person, meaning people would get the pronounciations wrong. If that's not a problem at encyclopedias, I don't know what is... --Kjoonlee 07:37, 11 August 2007 (UTC)
 * Now of course that's slightly problematic if people want to pronounce the words precisely. But I think the point you're missing is the purpose of an ad hoc system. Or, rather, its audience.  It would be in addition to the IPA transcription to give a sort of vague guide as to how the word's pronounced. It's not supposed to provide as much information as the IPA does (if it did, why bother?).  The latter is fine-grained, the former coarse: rough and ready. So you might have, for example: 셈 (pronounced, approximately "sem") and 샘 (pronounced , approximately "sem").  There is a significant proportion of readers who don't understand the IPA and can't be bothered to spend any time learning.  This page, you'll have noticed, is full of their whining.  But at least they can come away knowing that 셈 and 샘 are pronounced something a bit like "sem".  Here's the rub: even if they bothered looking up how to pronounce them properly, they might well still say "sem".  So this ad hoc thing would partly be a concession to ignorance, I admit, but I also have in mind people like my father, who have no talent for imitating foreign sounds, however hard they try.  For him, sem and sem – or maybe sem and seym – is about as good as it gets.  To put it another way, plenty of people don't want to know how to pronounce the words in Korean.  What they want is the nearest English approximation to the sound, and that is what an ad hoc guide would provide. garik 09:15, 11 August 2007 (UTC)
 * 셈 and 샘 in Korean are as different as "pig" and "big." Wikipedia is an encyclopedia, and thus inaccurate and approximate info should be avoided, especially when they are misleading. --Kjoonlee 09:42, 11 August 2007 (UTC)
 * If people need ad-hoc pronunciation, they shouldn't be looking in an encyclopedia. --Kjoonlee 09:45, 11 August 2007 (UTC)
 * I understand that they're as different as pig and big are from each other. But I wouldn't mind especially if, say, Arabic Wikipedia told people that "pig farm" is pronounced "approximately big farm", provided it also gave an accurate IPA-based guide.  But I agree that concessions to ignorance are somewhat regrettable.  In my heart, I'd certainly prefer it if people just stopped whining and just learnt how to read IPA transcriptions – and it really wouldn't take long to learn what they need!  But I know a lot of people aren't going to.   This seemed a reasonable compromise. garik 09:56, 11 August 2007 (UTC)

Underlined
IPA is still underlined on mouseover.SuperElephant 17:43, 20 August 2007 (UTC)
 * True, but we should supply the IPA class together with the link:
 * ʃ|
 * So that underlines are still suppressed on mouseover. You can choose from the above, depending on what sort of alttext you want. --Kjoonlee 18:00, 20 August 2007 (UTC)
 * So that underlines are still suppressed on mouseover. You can choose from the above, depending on what sort of alttext you want. --Kjoonlee 18:00, 20 August 2007 (UTC)

Need Template:IPA-request
There exists


 * "Cleanup-IPA, a request for inclusion of IPA pronunciation."

but let's take the article Chauvanism, which doesn't have any pronunciation noted. There ought to be a template that says
 * This article lacks IPA pronunciation information. A requset has been registered. You can go ahead and add it too, see ...

Without saying if the article has/has not other pronunciation information already in place.

OK, I just added SHOW-vin-ism, and the above request template. Jeez. Jidanni 17:12, 4 September 2007 (UTC)

Invariance
"Pronunciation in Wikipedia should be given in a way such that phonemic interpretations will not differ depending on the reader's regional dialect." goes the new version of this page. Technically this is incorrect. The way a Londoner would pronounce New York is completely different to how a Californian would pronounce it. The difference is not simply down to allophonic variation: they use different phonemes. Indeed there is no one set of phonemes for English; for example; PALM, LOT and THOUGHT ; represent three different vowels for me but not for Americans. Jɪmp 07:21, 5 September 2007 (UTC)
 * What that entails then is that they need to be transcribed differently, but in such a way that the Usonian sees them as containing the same vowel. We can't have a one-to-one match between symbol and phoneme. Are we going to accommodate Scottish, though? kwami 15:10, 6 September 2007 (UTC)
 * There are (sometimes) other work arounds, e.g., (SAMPA) /n(j)u:.jO:(r)k/ (let the reader pick which consonant to include according to his dialect) & /lQt/ (for some of us /A:/≠/Q/≠/O:/, for others /A:/=/Q/≠/O:/ or /A:/≠/Q/=/O:/, and for others still /A:/=/Q/=/O:/ ... let the reader decide what each one means, refering to some key where necessary). If we've got a Scottish editor who wants to include his pronunciation, why not accommodate? Jɪmp 17:17, 6 September 2007 (UTC)
 * The pronunciation respelling key was one such work around, but people who like the IPA get really upset by it. SAMPA is no different really from the IPA, so any workarounds that work for one should work for the other. Of course Id (sorry, FireFox doesnt like apostrophes) to see Scots included, just didnt know if that was practical. kwami 18:43, 6 September 2007 (UTC)
 * Of course, there are instances where dual transcriptions may be prefereable (e.g. for bath, cloth, etc.) or even a practical necessity (e.g. for lieutenant, schedule, etc.) Jɪmp 02:59, 7 September 2007 (UTC)

Broad IPA for English
The discussion about what IPA symbols to use for transcriptions in wikipedia seems to have died down. The manual of style just advises to use a fairly broad transcription. Overcoming the diversity of English dialects, a certain degree of compromise would be needed to come to uniform system. I have collected the data from the article International Phonetic Alphabet for English in a table and added a column for a proposed broad transcription advice. The proposed broad transcription has a nice and simple structure and should be easy to apply, while still giving quite a good impression on "international" English pronunciation. Of course, this cannot cover the case where words have phonemically different realisations in some dialects (tomato, etc), for which more than verson should be given.

How do do you think about adding this to the guideline? &minus;Woodstone 17:14, 14 October 2007 (UTC)


 * I speak something very close to standard US English, and the US isn't as distinct as this table makes it look, so I'm modifying a few vowels in that column, adding an extra row for the other schwa, and a few suggestions in the last column. For example, since ar is rolled in some dialects, and the approximant value is never distinctive, I see no need for the symbol <ɹ>; most broad transcriptions simply use . When a vowel differs in both quality and length, redundantly indicating both will make the transcription more legible (thus /ɒ/). For consistancy, I think it's best to stick to lax vowels before /r/ (thus /ɪr/ rather than /ir/) and diphthongs for both e: and o: (thus /oʊ/ rather than /oː/).
 * (I proposed to use the inverted r symbol for rhotacised vowels, to distinguish it from the normal r (by convention used for the consonant r, even though it is mostly not a trill in English); using lax vowels before r, full redundancy in length and quality, and diphthongs for long e and o are fine. &minus;Woodstone 10:23, 15 October 2007 (UTC)).
 * Note that there are quite a few distinctions which are not included here, such as /ær, ʌr, ɒr/, plus those who distinguish fern-fir-fur - do we just ignore these distinctions?
 * (we can add more rhotacized distinctions; I propose to ignore fine details like in fern-fir-fur. &minus;Woodstone 10:23, 15 October 2007 (UTC))
 * With the changes I made (except for omitting the length signs in /ɔː/ and /ɑː/), this is what we already have in the astronomy articles. I think it would be good to have a single guideline. kwami 02:15, 15 October 2007 (UTC)
 * P.S. We shouldn't mark aspiration. If we write tie as and die as, then sty would have to be  or , either of which would spark edit wars. kwami 02:51, 15 October 2007 (UTC)
 * P.P.S. I think we should encourage the use of the IPAEng template, and redirect the link to a much simpler key that would simply give sample words for each IPA symbol. We don't need all the complexity of the current article; we could have links for that at the end for those interested in following up. kwami 03:04, 15 October 2007 (UTC)
 * Other than a quick intro explaining that not all symbols will be distinct for all people, and some links at the end, I imagine a page something like this, plus consonants:


 * {| class='IPA wikitable'

! IPA !! example ! colspan="2" | full vowels ! colspan="2" | diphthongs ! colspan="2" | rhotacised vowels (/r/ silent in many areas) ! colspan="2" | reduced vowels
 * /ɪ/ || bid, bit
 * /iː/ || bead, beat
 * /ɛ/ || bed, bet
 * /æː/ || bad
 * /æ/ || bat
 * /ɒ/ || bod, pot
 * /ɑː/ || bahed, father
 * /ʌ/ || bud, but
 * /ɔː/ || bawd, bought
 * /ʊ/ || good, foot
 * /uː/ || booed, food
 * /ɒ/ || bod, pot
 * /ɑː/ || bahed, father
 * /ʌ/ || bud, but
 * /ɔː/ || bawd, bought
 * /ʊ/ || good, foot
 * /uː/ || booed, food
 * /ɔː/ || bawd, bought
 * /ʊ/ || good, foot
 * /uː/ || booed, food
 * /uː/ || booed, food
 * /uː/ || booed, food
 * /eɪ/ || bay, hay
 * /oʊ/ || beau, hoe
 * /ɔɪ/ || boy, hoy
 * /аɪ/ || buy, hi
 * /aʊ/ || bough, how
 * /ɔɪ/ || boy, hoy
 * /аɪ/ || buy, hi
 * /aʊ/ || bough, how
 * /aʊ/ || bough, how
 * /aʊ/ || bough, how
 * /ɜr/ || bird, myrrh
 * /ʌr/ || hurry
 * /ɪr/ || mirror
 * /ɪər/ || beer, mere
 * /ɛr/ || berry, merry
 * /ɛər/ || bear, mare
 * /ær/ || barrow, marry
 * /ɑr/ || bar, mar
 * /ɒr/ || moral, forage
 * /ɔr/ || born, for
 * /ɔər/ || boar, four, more
 * /ʊər/ || boor, moor
 * /ær/ || barrow, marry
 * /ɑr/ || bar, mar
 * /ɒr/ || moral, forage
 * /ɔr/ || born, for
 * /ɔər/ || boar, four, more
 * /ʊər/ || boor, moor
 * /ɔr/ || born, for
 * /ɔər/ || boar, four, more
 * /ʊər/ || boor, moor
 * /ɔər/ || boar, four, more
 * /ʊər/ || boor, moor
 * /ʊər/ || boor, moor
 * /ɨ/ || roses
 * /ə/ || Rosa’s
 * /ər/ || runners
 * /əl/ || bottle
 * /ən/ || button
 * /əm/ || rhythm
 * }
 * /əl/ || bottle
 * /ən/ || button
 * /əm/ || rhythm
 * }
 * /əm/ || rhythm
 * }


 * It would be nice to include other vowels before r, but I expect it would become so confused, with so many of our editors merging the vowels and making mistakes in transcription, that it would be best not to pretend to be any more precise than this. kwami 03:40, 15 October 2007 (UTC)


 * The use of /ɛr/ for bear and /ɪr/ for beer effectively precludes making the merry/Mary and Sirius/serious (Oxford transcription of RP: /ɛ/ vs. /ɛː/ and /ɪ/ vs. /ɪə/) distinctions, as in both cases the symbol chosen represents the first mentioned vowel but the example word contains the second. Merging these vowels (which are distinct not only in RP but in virtually all non-US varieties) strikes me as excessively US-centric.  I would propose using the "tense" symbols /er/ and /ir/ for bear and bear: problematic in some respects, but at least it would preserve the distinction and doesn't introduce further unfamiliar symbols.  Another idea would be to use the RP symbols with a following /r/: /bɪər/ etc.--JHJ 08:47, 15 October 2007 (UTC)


 * How about /mærɪ/ for Mary? And /i/ in serious?


 * Assuming you're still using /æ/ in mat (and hence marry) no, that would collapse the marry/Mary distinction instead. If this is going to make sense from a British (not just RP) point of view, it needs three separate symbols, one for marry/mat, one for merry/met and one for Mary/mare. (In Oxford transcription, those are /a/, /ɛ/ and /ɛː/ respectively.  In Wells's transcription, they're /æ/, /e/ and /eə/.)--JHJ 11:09, 15 October 2007 (UTC)


 * How's this, then? As I said above, I wanted to add more distinctions, but was afraid it would just cause confusion. I think we might want to also add the schwas, because people are going to make mistakes. If the choice is /ɪr/ vs. /iər/, and someone writes just /ir/, then it will be clear that the distinction was lost. kwami 14:11, 15 October 2007 (UTC)


 * At a glance that looks better. I'm not actually too bothered about the actual symbols used (given the dialect variation, no choice is going to work for everyone), only that Wikipedia doesn't end up with a standard system that's too closely based on non-East Coast American English and so confusing to everyone else.--JHJ 16:21, 15 October 2007 (UTC)

(lose indent) Now I think you're overdoing it. Let's not go into too much detail in possible phoneme splits. First, we should not indicate in this table the intervocal r's, because they are realised in all dialects and can be derived by considering them as initial of the second syllable. Using /e/ for "bear" strikes me as odd. Splitting "four" from "for" seems unnecessary. I edited the table at the top of this section to reflect a new proposal. &minus;Woodstone 16:25, 15 October 2007 (UTC) With this: &minus;Woodstone 16:46, 15 October 2007 (UTC)
 * Mary = /mæːrɪ/, mare = /mæɹ/ (lengthening absorbed by rhotacising)
 * Marry = /mærɪ/, mat = /mæt/
 * Merry = /mɛrɪ/, met = /mɛt/


 * I don't agree. All of these are distinctive in RP, as far as I can tell (I don't control that dialect), so I don't see why this should be overdoing it. It's not like I'm trying to recover the fir-fur-fern merger. /er/ for "bear" is analogous to the other vowels. (/ɛər/ or /eər/ would be okay, but only if we change the others.) Also, no, the intervocalic ars are not realized in all dialects; they're merged as well, so we need to give examples for people who don't distinguish them. You stated in your edit summary that you'd like to use ar /r/ when intervocalic, but turned ar /ɹ/ elsewhere; however, this is an allophonic distinction (all of your turned ars would turn right-side up again when there's a following vowel-initial word, for example), so it's not appropriate for a broad transcription. kwami 16:58, 15 October 2007 (UTC)


 * Isn't the distinction bar = /bɑr/ and moral = /mɒrəl/ allophonic as well? &minus;Woodstone
 * I don't think so. You have marrer /mɑrɚ/ in RP, and in my dialect marl is /mɑrəl/, whereas moral is /mɔrəl/; for me it's the same vowel as in boar, not bar. kwami
 * One of the biggest dialectic differences is the r after a vowel. We can use one symbol for it and state that it is not realised in some dialects.
 * Yes.
 * Intervocalic r can be seen as initial of the second syllable.
 * Yes.
 * Can you give some examples where intervocalic r is not realised?
 * My point was that in many dialects vowels which are otherwise distinct collapse before /r/. For example, in my dialect met is distinct from mat, but I don't distinguish merry from marry. Therefore I think we need to separately treat all simple vowels that occur before /r/; we can probably ignore diphthongs like flower and fire.
 * That on the other hand final r is realised before a following word starting with a vowel is clear, but how is that hindered by the use of an inverted r (which is the more correct symbol anyway).
 * We can use whichever symbol we choose, but  is standard for broad English transcription. Even the OED, which uses RP, has . Whichever symbol we choose, we need to stick to it, because there is only one /r/ in English; with your transcription, you'd be saying that the r of mar is a different phoneme than the mid r of marrer.
 * For the rhotacised vowels, other editors (above) suggested to always use the lax variant (that's why /ɛ/, not /e/ in "bear" and /ɪ/, not /i/ in "beer"). For the simple vowels is chosen for redundancy between length and quality (/ɪ/ vs /iː/ etc.).
 * I was thinking the redundancy would help avoid confusion, but after thinking about it I see it probably doesn't. I now agree that we should stick to lax vowels, and have changed them in my table.
 * The vowel /æ/ does not have a lax variant, so there redundancy os lost and it is only short or long. On itself I have no objection against /ɛər/, but it is not so clean for stating the difference between US and UK.
 * Not so clean as what, <ɛːr>? My objection to that convention is that it becomes the only place that the length sign <ː> is phonemic; everywhere else our readers can ignore either length or the tense-lax distinction, which I think will be helpful to those who are not familiar with the IPA. It also makes for a more symmetrical inventory, again helpful to those who need to consult a key. As for Mary, I think that's already covered.
 * I hope that with this proposal we can also document the basic principles used to make the compromise choices, to come to a balanced, regular system, not too far from most dialects and sticking to broad interpretation guidelines of the IPA. &minus;Woodstone 17:25, 15 October 2007 (UTC)
 * Agreed. kwami 19:20, 15 October 2007 (UTC)


 * I have added (in your table) some example words with slightly shorter vowels. For consistency I have replaced the combined rhotacized symbols by an explicit r. And I have added the sonantic liquids, to which you have not commented yet. What is "bahed"? &minus;Woodstone 19:56, 15 October 2007 (UTC)
 * Oh, bahed is just the past tense of (to say) bah. A bit weird, but I couldn't think of anything else that fit our b-d set. We should probably drop it. As for writing all the ars explicitly, do they all behave the same? That is, is RP for runner-up [ɹʷɐnəɹʷəp], similar to run a wrap? I wasn't sure (it's not given in the OED), which is why I left ɝ and ɚ. What you have is closer to the OED, but ɚ and ɝ are very common, so maybe we should list both? kwami 20:15, 15 October 2007 (UTC)
 * Here's what I had in mind to link to the template: Help:IPA English pronunciation key. kwami 20:24, 15 October 2007 (UTC)


 * The total of the proposal is more coherent if we use /r/ throughout, no rhotic diacritics (not any diacritics actually). Your example I would distinguish by keeping /ʌ/ or /æ/ in the last syllable (just a slightly more cultivated form). Vowel reduction is a matter of degree, not an absolute. How do we proceed from here? Get more opinions? Insert it in the Guideline? Of course we should add a similar chart for the consonants (easy to do). &minus;Woodstone 20:30, 15 October 2007 (UTC)


 * Had a look at the help page. Looks good. However, I'm for having a definite choice everwhere, no alternatives. No diacritics at all. Nor the (AuE?) /əʊ/. The cleaner end straighter it is, the easier it will be to follow. I think we need long and short /æ/ (e.g for mary/marry) (no redundancy here, no lax symbol available). How about adding the sonantic liquids? Do you want to keep the dialectic realisation chart somewhere around as supporting material? To show the balance of the choices? &minus;Woodstone 20:37, 15 October 2007 (UTC)


 * Okay, let's get rid of the /əʊ/ (actually, that's the pronunciation of all three dialects, RP, GA, and Oz, but /oʊ/ will be more intuitive to our readers). As for ɝ and ɚ, the proplem is that they're already found in hundreds of articles, so I'd like to keep them, maybe in parentheses. (Whereas /əʊ/ vs. /oʊ/ encode different phones, you could argue /ər/ and /ɚ/ are just different symbols for the same thing.) I didn't add /əl, əm, ən/ because I thought they were self evident when read. Remember, this is a key for the reader, not a style guide for the editor, so we don't need to specify how to transcribe bottle. (Okay, I added them in as a note.) As for Mary/marry, per the OED that's already covered by /ɛər/ vs. /ær/, vs. /ɛr/ for merry. kwami 22:13, 15 October 2007 (UTC)
 * Oh, and I'd like to keep dialectal differences out of this. There's a link for that at the bottom of the page, but if we put it in (a) we lose the simplicity of a pronunciation key, and (b) I'm afraid we'll start getting into edit wars on adding every reader's dialect, the proper transcription of the dialects we do have, etc. Better to leave that for an article dedicated to the subject. kwami 22:17, 15 October 2007 (UTC)

Linked it to the template. Even if we modify the table, as it is now it should still be more accessible to our readers. kwami 22:34, 15 October 2007 (UTC)


 * I see the list both as a guideline (in the style manual) how to create broad international English IPA transcriptions as well as a guide for the reader how to interpret them. So I still think we should add the /əl, əm, ən/, but I agree with the remainder of your comments. &minus;Woodstone 09:18, 16 October 2007 (UTC)
 * They're currently in there, in the notes. Hopefully, anyone using this as a style guide will take the time to read it, but they're out of the way of our readers. kwami 10:17, 16 October 2007 (UTC)
 * Also, you might want to take a look at some criticism on the template Talk page. kwami 10:18, 16 October 2007 (UTC)

Chart for all IPA characters
There was a discussion at the end of September 2007 at "Talk:IPA chart for English" which concluded with the consensus that there needs to be a chart intelligible to readers who do not know IPA that explains how all IPA symbols are pronounced. There is little point in insisting on inserting IPA transcriptions into articles (which I have nothing against) if there is no key to help the ordinary reader interpret the symbols. Has someone got cracking on this yet? Cheers, Jacklee 02:44, 18 October 2007 (UTC)


 * IPA-en now links to a basic pronunciation key that covers nondialect-specific English only. A more complex key on a different help page could be linked to IPA2, but explanations of the sounds would be difficult. kwami 03:36, 18 October 2007 (UTC)
 * I'm starting a page here: Help:IPA pronunciation key. I may or may not get around to doing it myself. Once it's up and running, we can redirect IPA2. kwami 03:48, 18 October 2007 (UTC)

Awesome! Start with the symbols (including diphthongs) used in English first, and then move on to the symbols that are only used in other languages. Despite the difficulty of explaining the latter, I think an effort still has to be made to describe every IPA symbol. I reiterate that it's not much point inserting IPA transcriptions into articles (and removing respelled pronunciations, as you did for articles such as "Ulrich Mühe") if ordinary readers cannot find out how the symbols are pronounced. To explain foreign sounds, I think that the key will have to use a combination of commonly-encountered foreign words, and descriptions (eg, "sounds like 'oo' with pursed lips"; "sounds approximately like 'ch'"). Cheers, Jacklee 04:17, 18 October 2007 (UTC)


 * See how it looks now. I don't want to organize by which language a phone is found in, because that will make it very difficult for our readers to look things up. Also, for English there's the English IPA key. This chart can be linked to the IPA2 template that's used primarily to transcribe foreign words, so there's no point in putting English first. (I envisage people using IPA-en for a broad transcription of English words, and IPA2 for a phonetic transcription of foreign words.) Also, some symbols are so rare that they will have a paragraph devoted to them in any article that contains them, so they'd be too much here. I need more examples, especially from German and Hindi, but I think it's best to keep things simple so that people aren't overwhelmed. Once the symbols are linked to their individual articles, people can continue on if our descriptions aren't enough for them. kwami 09:14, 18 October 2007 (UTC)

Thanks Kwami, the work you've done on "Help:IPA pronunciation key" is really excellent. I'd suggest that you standardize the table sizes and formalize the language ("there's" to "there is", for instance), but otherwise it's exactly what we need in Wikipedia. I agree that very rare symbols can be omitted for now, and added if there is a request for it. You might post messages at "WikiProject Germany" and "WikiProject India" to see if you can get assistance from German- and Hindi-speakers with examples. I had a look at IPA2 – I assume you'll be writing detailed documentation on when we should use IPA, IPA2, IPA3, IPA-en and other variants? Cheers, Jacklee 13:45, 18 October 2007 (UTC)


 * Where would such documentation go? I've put Usage sections on the templates themselves, if that helps. A lot of people are working on the chart right now, so there will be formatting issues until things settle down. Right now, though, we need someone to help with the links. Once that's done, we can link to the IPA2 template. kwami 21:06, 18 October 2007 (UTC)

Well, I was thinking of some documentation on the templates themselves that explained what all the different IPA templates are for. You could do a single documentation page at, say, "Template talk:IPA/doc" and transclude that on to the other template pages. In that way, it's possible to see at one glance which is the right template to use. I did this for Malay name. What links do you need help with? Cheers, Jacklee 22:07, 18 October 2007 (UTC)
 * You might want to join the discussion here, so that we can iron things out first. kwami 23:59, 18 October 2007 (UTC)
 * Okay, I need to create another template for this to work right. kwami 23:32, 21 October 2007 (UTC)

Chiodos
A while ago, I changed the pronunciation at the article Chiodos to an IPA transcription of the pronunciation provided in the source for it. (For the record, said source is an interview with the lead singer.) Since that time, a slew of anonymous IPs have changed the spelling to ad hoc American pronunciations which I think run afoul of the need to give pronunciations "in a way such that phonemic interpretations will not differ depending on the reader's regional dialect". (A small contingent of IPs also insists on changing the pronunciation to something that diverges from the source, but that's neither here nor there, because they have ignored all requests for a source for that pronunciation.)

See this diff for an example.

I have reverted these changes. But am I totally off base here? Should I just let it go? It seems like it really should be either in IPA or in some other recognized phonetic transcription, but am I wrong in pushing this point?

Thanks. - Revolving Bugbear  (formerly Che Nuevara) 18:32, 5 November 2007 (UTC)
 * No, you're not. I have no problem with pronunciation respellings, but they should be an adjunct to the IPA, never replace it, and they should follow a standardized system, as at Help:Pronunciation respelling key. But where is the stress? kwami 19:12, 5 November 2007 (UTC)


 * Thanks for the reassurance.
 * I actually don't know where the stress goes. I've never heard of the band outside of Wikipedia myself, it just wound up on my watchlist after an AfD. - Revolving Bugbear  (formerly Che Nuevara) 19:14, 5 November 2007 (UTC)


 * I feel that approximate respelled pronunciations, particularly for foreign terms, according to "Help:Pronunciation respelling key" are helpful when used in conjunction with IPA transcriptions, but whenever I've inserted respellings into articles, even with a footnote explaining that this is just an approximate pronunciation, they have been deleted by other editors saying that they are inaccurate or confusing. See, for instance, the discussion at "Talk: Garðar Thór Cortes". It may be helpful to start a discussion on this talk page about the matter with the aim of discovering if there is consensus for this Manual of Style guideline to permit the use of respelled pronunciations. However, I do note that there may be less need for respelled pronunciations now that we have articles explaining the meaning of IPA symbols. &mdash; Cheers, Jack  Lee  –talk– 21:48, 5 November 2007 (UTC)


 * It's a very partisan issue. One side says that respellings are inferior even for English (which isn't true), and that everyone should learn the IPA; the other says that many people don't know the IPA, and that many people are only looking for a general idea as to the pronunciation. I've tried discussing this in the past, and people really get worked up even without me trying to approximate foreign words. Maybe for a foreign name we could use the IPA followed by approximately X with a respelling, but I bet you'd need to keep them all on your watch page. kwami 04:12, 6 November 2007 (UTC)
 * BTW, it tends to be the editors to the Manual of Style guideline who insist on the IPA, and readers (non-editors) who complain that it's unintelligible. (And not just Usonians either.) So discussion on the Manual of Style page might be somewhat unrepresentative of what our readers want. kwami 04:17, 6 November 2007 (UTC)


 * If the issue is likely to invoke the snowball clause, then it's not worth starting a debate on it. But if you think there's a chance that some consensus or compromise can be reached, perhaps it's worth trying. This talk page would seem to be the logical place for the debate, since any change would have to be to this portion of the Manual of Style. The question is how to publicize the debate widely enough so a number of editors, both those in support of and against respelled pronunciations, can choose whether or not to participate in it. We're currently involved in a debate over at "Manual of Style (biographies)" over when to use "British" and "English/(Northern) Irish/Scottish/Welsh" to describe persons, and notice of the discussion was given to the WikiProjects of Britain, England, Northern Ireland, Scotland and Wales. But in this case I'm not sure what's the best way of publicizing the matter. Any ideas? &mdash; Cheers, Jack Lee  –talk– 05:27, 7 November 2007 (UTC)
 * The respelling key was really intended for English only. If we try using it for foreign words, we're going to have to ignore a lot of distinctions. Someone will almost always object that we ignored them incorrectly, or that it's wrong to ignore them in the first place. For instance, we'd have to ignore the voiceless ar in your example. Also, in Icelandic, would initial b be written b or p? Either way you're going to get an argument. b is phonetically incorrect, since Icelandic b is not voiced, but then p is also wrong, and further lumps it together with Icelandic p. BTW, I redirected that article to a simpler IPA link. kwami 06:21, 7 November 2007 (UTC)

phonetic vs. phonemic
I want to open this up for discussion.

After hearing people complain about the IPA and pronunciation respellings for years now, and not seeing the situation get much better, I've tried over the last few weeks to bring some order to pronunciation in Wikipedia. I've reduced the number of links to the unwieldy main IPA article from 17,000 to 3500, redirecting them to the new help keys and often copy editing them for consistency. I've had a few vocal critics, however, and although I may have been short with them at times, they've brought up some important issues, and I think we don't need to be in conflict as long as we make it clear what we're doing. (A lot of the pronunciations I haven't got to are still ambiguous or utterly unintelligible, and I don't want to go back to chaos. And yes, I know I've introduced my share of errors.)

This was the simple (or perhaps simplistic) vision I had:
 * English pronunciations should be in a non-dialect dependent transcription, between /slashes/ and linked to a key which explains the IPA symbols in terms of English key words, so that our readers don't need to know anything about articulation, phonation, and the like, or about the dialect (RP, GA, etc.) of the editor.
 * Foreign pronunciations, on the other hand, should be broadly phonetic, between [brackets], and linked to a more general IPA key, so that our readers don't need to know anything about the phonology of the language in question in order to interpret it.

However, a couple editors want phonetic transcriptions for some English personal and place names, and phonemic transcriptions for some foreign names, especially French and Irish ones. The problem with the former is that most people won't know how to pronounce the names in their own dialect, and the problem with the latter is that a phonemic transcription is an abstraction; it doesn't tell you how to actually pronounce anything.

I have nothing against either, given a few precautions: Does this sound like the way to go, something we should adopt for the MoS? kwami (talk) 09:52, 18 November 2007 (UTC)
 * 1) People should be able to assume that an English transcription is targeted to the English-speaking world in general, regardless of which dialect they speak. That is, for a basic entry, York should be with an R, and New York should be  with a J. If an English pronunciation is anything other than generic, whether dialect specific, the local pronunciation of a town, how someone pronounces their own name, or historical (say from sound recordings or a reconstruction from poetry), then we should state this explicitly: "locally [X]", or "[Y] in RP", or "she pronounced her name [Z]". (Either phonetic or phonemic transcriptions would work for any of these.)
 * 2) Foreign names should only be transcribed /phonemically/ if they link to an article on the phonology of the language in question, or ideally to an IPA help key for that language, so that they can be interpreted correctly. TimeInEurope has done this for some Irish pronunciations, and someone else has done it for Italian. Otherwise we should stick to [bracket] notation. I think if we're going to go the phonemic route, we should use language-specific IPA templates, as some do for Australian English. That way all such entries can be tracked and kept on a standard footing. (We don't want editors with opposing theories about the phonology of a language creating contradictory transcriptions, a problem I've encountered with English, where /e/ has sometimes been our, sometimes our , and sometimes undecipherable.) And if, say, a phonology article is split off from the main language article, or something like International Phonetic Alphabet for Irish from the phonology article, all the links can be changed without tracking down and editing the articles individually. Believe me, doing this for the main IPA article has been a royal pain: My ears are literally ringing right now from the odd hours I've been keeping. Also, if you don't know how to do make a template (though it's pretty easy), I and I'm sure many other editors will be happy to create IPAIr, pronIr, IPAFr, pronFr, or the like, and link them to any article you want.

Referencing
Based on recent experience, the accepted practice for inserting pronunciation into an article seems to be either one of the following:


 * 1) An editor can insert an IPA phonetic entry from the Oxford or Cambridge dictionaries without change.
 * 2) An editor can use any dictionary and manually convert the phonetics to IPA.

In either case, I have run into some resistance when pressing for a reference. (I.e. a fact gets almost immediately reverted, even though this conflicts with Citing sources for material that is challenged.) Hence I would greatly appreciate it if this document could formerly state that the second is acceptible practice (despite my WP:OR reservations) and that no reference is required unless there is controversy over the pronunciation. That would generally alleviate some concerns regarding these entries.

Thank you!&mdash;RJH (talk) 22:53, 19 February 2008 (UTC)


 * 1) I've been deleting some of those. Fact tags on words like 'Mars' that can be found in any dictionary are in my opinion excessive. (And they have seldom been used, even in featured articles.) There are plenty of fact tags on personal or place names where it truly isn't clear that the transcription is accurate, some of which I put up months ago and which are still there.
 * 2) Since you will seldom get two sources that use the same IPA conventions for transcribing English (the 2nd and 3rd editions of the OED don't even agree), there will have to be some conversion, unless we agree to only use a specific edition of a specific dictionary and that anyone who doesn't have access to that edition can't contribute—and even then we won't be able to provide most of the personal or place names that people most need to look up. We certainly can't afford to use a dozen IPA keys to stay true to a dozen source dictionaries, when many of our readers have trouble with just one. kwami (talk) 01:45, 20 February 2008 (UTC)

Wikipedia should establish a single IPA transcription policy as an in-house standard. Any other policy would be unworkable. As long as Wikipedia's IPA page links to a singe established outside authority, there shouldn't be a problem. The only issue comes with what pronunciations actually are, not with how they are transcribed. So for instance, the Kuiper belt page links to an outside source which establishes that the proper way to pronounce "Kuiper" is to rhyme with "viper", not "Koy-per." How the IPA is used to transcribe this pronunciation should be based only on Wikipedia in-house rules.  Serendi pod ous  09:23, 20 February 2008 (UTC)


 * Yes, basically I would like to see a single policy written up on how the pronunciation inserts should be generated. When is a citation required (or not) and what sources are suitable? Rather than depending on various opinions that change from person to person, I'd like to see it in writing as a consensus MoS policy that I can point to during discussions. Right now there seems to be little consistency and everybody is just following their own preferences. The result is contention, confusion, frustration and edit wars.&mdash;RJH (talk) 21:19, 20 February 2008 (UTC)


 * I have one recommendation that I think is important, though not the point of this discussion: If we give a phonemic transcription of a foreign word, we must link to an explanation, such as the phonology page of the language in question, and it must use the same conventions, so that our readers can follow the transcription. I think the default for foreign words should instead be a phonetic transcription, so that we can use any source, and that phonemic transcriptions should only be used for the few languages for which we have numerous competent editors, such as French, Irish, etc. Since there is little theoretical presumption in phonetic transcriptions, we don't need to establish in-house transcription rules. (For English, of course, we do need such rules.)
 * I don't know where we draw the line, but it does seem to be a little excessive to demand sources for English words that can be found in any grammar-school dictionary. We could have a common-sense approach, as we do now for when to add pronunciations at all. However, I wouldn't oppose such a requirement.
 * (But let's not fact tag such entries unless they're up for GA or FA, okay? That could quickly become a real mess.)
 * Well, if that's to be the consensus (and so far it's a consensus of one), then I'd like to see the lack of necessity for a citation written up in the MoS. Otherwise I'll continue to take issue with people reverting pronunciation entries that I got from a legitimate online dictionary, then providing no source to back up their own version.&mdash;RJH (talk) 16:16, 21 February 2008 (UTC)
 * The Help:Pronunciation page contains in-house rules that are something of a compromise: OED vowels (with the exception of replacing non-IPA <> with IPA <>, and <> with <>), but retaining GA ars and aitches. This is a tough sell for place names like New York, where people object to the RP vowel even though it's unambiguous, but over all it's been well accepted: Brits don't seem to have a problem adding ars to place names where they aren't pronounced, for example, though it helps if we use the less obvious symbols <> and <> in such cases. It's very similar to the transcription of dictionary.com, except that 'syllabic' el, em, en are written with a schwa rather than just the consonant. (Writing just the consonant becomes problematic in many words where people can't agree, but I don't believe this is a phonemic difference.)
 * As for which sources we use, that is tricky. Many American dictionaries do not make all the vowel distinctions of RP, which means that if we use them we exclude half our readership. In the case of place names, there may be no agreed-upon RP pronunciation, so we either exclude them, make an educated guess based on spelling pronunciations, or add the awkward statement that we don't know how to pronounce these names even though we give a source. (Either that or include only the local pronunciation, which is probably the way we should go in such cases.)
 * kwami (talk) 22:08, 20 February 2008 (UTC)
 * It seems like the wiktionary would be the best location for the pronunciation information. Perhaps the fine details and local variations of pronunciation could be covered there, along with references if necessary, then a brief pronunciation entry on wikipedia could just be trans-wikilinked over to the wiktionary entry?&mdash;RJH (talk) 16:22, 21 February 2008 (UTC)
 * That could work. It'd be convenient to keep the basics here in Wikipedia for semantic categories where people look up lots of names they don't know how to pronounce - dinosaurs, for instance, or the names of Saturn's moons. Or words which lots of people get wrong, like boustrophedon. And we could still keep track of vandalism or ignorant edits, since that's not likely to be a problem in Wiktionary. But do we even have Wiktionary entries for personal and place names? kwami (talk) 23:55, 21 February 2008 (UTC)
 * I easily found a number of personal and place names in the wiktionary, so it seems to be accepted practice. Thanks.&mdash;RJH (talk) 15:52, 3 March 2008 (UTC)

IPA reader in Wikipedia
I'm wondering, why an IPA reader (A script that creates WAV/OGG file on the fly) can't be added in Wikipedia? Appending of the WAV files is easy -- Rrjanbiah (talk) 05:53, 20 March 2008 (UTC)

Opentail g
Should Wikipedia transcriptions of the voiced velar plosive use the ASCII-standard "Latin small letter G" (U+0067) or the IPA-preferred  "Latin small letter script G" (U+0261)? I ask because list of names in English with counterintuitive pronunciations is mixing both which confused me and which I think is undesirable. Maybe IPA and its derivatives could auto-substitute to whichever standard. jnestorius(talk) 10:35, 20 March 2008 (UTC)


 * The cause of mixing them is, that for most editors, there is no difference between them in appearance (both in the edit box and article view). Moreover, they are never distinctive in IPA. They are just two forms of the same symbol, both in standard international and English typesetting and in IPA. Therefore my preference is to use the normal "g" on the keyboard and leave display form to the browser. &minus;Woodstone (talk) 11:13, 20 March 2008 (UTC)
 * I agree. In my browser, ɡ shows up as a square in the edit box while g is, well, g.  When the IPA template is used, both are identical.  — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi]  22:55, 6 April 2008 (UTC)

Just wondering
I guess that you are probably very proud to use a correct and "scientific" transcription scheme to describe pronunciation. IPA is lovely indeed. But have you ever stoped and consider that a lay person cannot make any sense of the pronounciation *guide* that you provide? Sarah, 82.152.207.52 (talk) 17:49, 19 April 2008 (UTC)


 * Of course we have, but there's no practical alternative. We can't do much about people who don't know how to use a dictionary. kwami (talk) 17:57, 19 April 2008 (UTC)
 * The average middle-school student in Korea will be very familiar with the IPA symbols, because English dictionaries in Korea use an almost-identical set of symbols. The same goes for some other countries. The IPA is indeed International and there's no other choice for an international encyclopedia such as Wikipedia. --Kjoonlee 18:39, 19 April 2008 (UTC)
 * Kindly also note that help on the IPA has been steadily improving: we have Help:IPA and Help:Pronunciation. If you search for a specific symbol here at Wikipedia, you're very likely to find recordings for the sounds as well. --Kjoonlee 18:41, 19 April 2008 (UTC)
 * If you still find the guide unclear, kindly tell us how we can improve. If the guide is unclear for you, it's because you didn't care to help. --Kjoonlee 18:44, 19 April 2008 (UTC)
 * There certainly is another choice for Wikipedia, and that choice is English respelling pronunciations. The average reader who consults Wikipedia is an anglophone, and when he encounters a difficult English word rendered in IPA, his reaction is invariably to simply ignore it, which leaves him just as ignorant of its pronunciation as he was before.  This is a disservice to the reader.  For difficult, "hard," or obscure English words (and especially those which are regularly mispronounced by English-speakers), an English respelling pronunciation should always be given in addition to the IPA rendering.
 * Don't even dream that the American education system is ever going to teach the IPA in middle school! Morris K. (talk) 19:16, 19 April 2008 (UTC)
 * In that case you can help by adding them (in addition to the IPA) and/or providing sound recordings of words and names. Wikipedia does not exist to serve Americans. --Kjoonlee 19:30, 19 April 2008 (UTC)
 * Kindly also note that Wikipedia doesn't just list subjects familiar to the Anglosphere or to the Americas. These need to be described as well, and respelling systems tend to be inadequate for this job. --Kjoonlee 19:33, 19 April 2008 (UTC)

Kindly note that subjects unfamiliar to the anglosphere are precisely the ones for which English respelling pronunciations are most suited and most needed; they are quite adequate for the job if only they would be used. And, as you have said, the IPA rendering should be given in addition to them. Wikipedia does not exist just to serve trained phoneticians. Morris K. (talk) 19:53, 19 April 2008 (UTC)


 * They're not suited at all. If you look at the Help:IPA page there are lots of symbols not covered by ad-hoc styles. --Kjoonlee 12:08, 20 April 2008 (UTC)


 * Come on, people. Not only is this biting the new guys, but those questions were perfectly reasonable.  Remember WP:NOTLEX, which says, more or less, pay attention to the dictionaries.  The consensus of hundreds of thousands of people who are paid to write persuasively is probably going to be a better answer than whatever occurs to any of us off the tops of our heads.  In this case, the questioners have it right: almost all of the dictionaries most commonly used by native-English-speaking Wikipedians have decided that transcriptions such as "prĭ-kō'shəs" are easier for us to read than "/prəˈkəʊʃəs/".  Does anyone know of a wiki-tool or browser plug-in that will automatically convert IPA to something easier for most English speakers? - Dan Dank55 (talk) 19:59, 19 April 2008 (UTC)


 * P.S. I looked around a bit, can't find a useful conversion template or tool. Pronunciation_respelling_for_English and Help:Pronunciation respelling key are a good start.  Btw, I'm not disagreeing with this guide; IPA is better for an international readership.  But some kind of pronunciation tool would be in line with current dictionaries and very helpful. - Dan Dank55 (talk) 20:07, 19 April 2008 (UTC)


 * Since Wiktionary gives both IPA and something they call "enPR" (which is close to the transcription that American Heritage uses), I'm wondering if they use a tool for that, I'll go ask. - Dan Dank55 (talk) 20:42, 19 April 2008 (UTC)

This is really amusing. I came here because my sister pointed this to me. She does know how to read a dictonary (?!?!) and she does not understand IPA. She is English, not Korean (that probably explains it). Koon writes: "it's because you didn't care to help". Well I do care to help, that's precisely why I asked this question. I'm not so sure if I care anymore after what I read here. Have a nice fight and best wishes, Sarah —Preceding unsigned comment added by 82.152.207.52 (talk • contribs)
 * The point was that it's not helpful if you find it confusing but walk away rather than explain how you're confused. This is, of course, assuming that you were confused.  — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi]  02:04, 20 April 2008 (UTC)


 * My comment was inappropriate. I was woken in the early hours of the morning by a guy in Jamaica who called nine times (that I was awake enough to count) over the course of an hour, trying to "give" $100,000 to someone he refused to believe doesn't live here, couldn't get back to sleep, and have been too groggy to finish a paper I'd promised a lot of people I'd send off today. Sarah was the next person I communicated with, so I snapped at her. Sorry. kwami (talk) 03:16, 20 April 2008 (UTC)
 * If you and your sister live in England, then you probably have access to COBUILD dictionaries. They use IPA (or something very similar) AFAIK. --Kjoonlee 13:56, 20 April 2008 (UTC)
 * The OED is also going IPA, so it looks like the wave of the future in the UK. The US is a different matter. kwami (talk) 14:15, 20 April 2008 (UTC)

Rhymes with
User:Timeineurope seems intent changing the wording of the article thus to remove the permissability of "rhymes with" or "stress on the first syllable." I certainly disagree with this change and checking the archives shows a pretty clear consensus on on the matter.

TE, I don't know if you're being arrogant or ignorant here but it doesn't really matter. You shouldn't be editing policy pages like this. If you'd like to make changes to MOS you should get a consensus first. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi] 23:25, 20 April 2008 (UTC)
 * Actually, I'd say that bit goes against the whole spirit of the IPA and this policy. We shouldn't be using "rhymes with" or ad-hoc styles. --Kjoonlee 11:57, 21 April 2008 (UTC)
 * Rhymes with trap, rhymes with bath. Rhymes with cot, rhymes with caught. It should sound familiar. And chilling. --Kjoonlee 12:05, 21 April 2008 (UTC)
 * (from the archives)Wikipedia is for general consumption, it's not a specialist resource for linguists.1 It should be allowed for some cases to write  "stress on the second syllable" or "rhymes with...". Most often IPA will do just fine.2  The ad-hoc descriptive workarounds:3
 * are easy to understand
 * are only possible for some cases
 * may be verbose
 * may be accent-specific
 * (from me) Obviously the opportunities for using them are few, but this allows editors not familiar with the IPA to make attempts at indicating pronunciation and readers not familiar with the IPA to understand examples that would be laborious to understand even with pronunciation respelling. If someone uses this in a non-helpful way, we can refer to common sense in changing it to IPA. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi] 19:01, 21 April 2008 (UTC)
 * It's not a policy page, it's a guideline. And why did you also revert my implementation of the new consensus at Help talk:Pronunciation – which you yourself were a part of – that respellings need not conform to the respelling key?


 * I was actually not trying to 'remove the permissability' of anything. 'Rhymes with' would still be permissible even if the guideline didn't explicitly say so. In the edit summary, I wrote that explicit endorsement of 'rhymes with' is a bad idea, as most people have no idea which words rhyme in all dialects of English. Timeineurope (talk) 21:33, 21 April 2008 (UTC)

There's another reason to keep it: A lot of editors use the IPA incorrectly. Sometimes it's nearly impossible to tell what they meant, and a "rhymes with" or "stress on" can either disambiguate, or confirm that an odd pronunciation is not an error. For example, if stress is marked for the 2nd syllable, but the editor states it's on the first, there's a good chance they're using the stress mark after the stressed syllable as in Usonian dictionaries. Or they may transcribe a vowel /e/, and it's only the "sounds like" that tells you whether that's supposed to be /eɪ/ or /ɛ/. It's at least enough to warn us there's a problem, so we can bring it up on the article's talk page. kwami (talk) 20:02, 21 April 2008 (UTC)
 * I guess I'm guilty of throwing the baby out with the bathwater, TE. I apologize.  It seems sort of Orwellian to say that removing explicit permission of something from the MOS isn't removing its permissability.  How do we know it's permitted then? — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi]  22:03, 21 April 2008 (UTC)

Eh? I tweaked the manual again without noticing this whole discussion. (I only saw Aeusoes1's initial response to my comment.) I hope I did it in the right direction. --Kjoonlee 14:28, 23 April 2008 (UTC)

Mixing foreign and English pronunciations
Now TE is waging an edit war claiming that Pachelbel is pronounced in English, evidently because people who know German might use the German pronunciation while speaking English. Shoot, that's half the reason we include native pronunciations in the first place. I'm tired of squabbling with him. Any opinions from the rest of you? kwami (talk) 22:47, 21 April 2008 (UTC)
 * No, that's not what I'm claiming (and it takes two to wage an edit war). I'm saying that since some people will use a German pronunciation while speaking English, it's not correct to say that only the non-German pronunciations are used in English. Timeineurope (talk) 23:09, 21 April 2008 (UTC)
 * * sigh* Okay, not "people who know German", then, "people who know the German pronunciation". It's still not an English pronunciation. kwami (talk) 23:57, 21 April 2008 (UTC)
 * No, I take that back. Few English speakers would be likely to get the German vowels right. Pronouncing it not like German, but in German, is code switching. kwami (talk) 00:01, 22 April 2008 (UTC)
 * I'm saying the German pronunciations can be used in English sentences. You appear to agree with that. Without the 'also', people will think only the English-only pronunciations can be used in English sentences. Timeineurope (talk) 19:20, 22 April 2008 (UTC)


 * per WP:NOTLEX, and dictionary.com the English pronunciation should be and the German.
 * is OR and incorrect. I'm assuming since this is brought up here and not at Talk:Johann Pachelbel that the issue is about when English speakers try to pronounce foreign words.  It's called code-switching, people do it all the time in broadcast media ("a shooting occured today at a young girl's quinceañera.  Seven people, including community leader Gloria Angélica Moraga Rodriguez are in critical condition.") and such pronunciations shouldn't be considered English pronunciations, even when English speakers use them.
 * Timeineurope is somewhat new to Wikipedia. Kwami, maybe you should show him how not to edit war.  — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi]  01:03, 22 April 2008 (UTC)
 * Aeusoes1, how about looking into the matter before you comment on it? The pronunciation you label 'OR and incorrect' is cited in the article. It's one of three German pronunciations of Pachelbel from the Duden-Aussprachewörterbuch. The disagreement between Kwami and me is about the inclusion of the word 'also' before the two English-only pronunciations given. I think it should be there because otherwise people will interpret the article as saying that the three German pronunciations from the Duden-Aussprachewörterbuch can't be used in English sentences. Obviously they can, so the 'also' should be there to make that clear.


 * You do realise that an edit war takes at least two editors? Kwami is the last person who should be asked to show people how not to edit war, he loves them. Timeineurope (talk) 19:20, 22 April 2008 (UTC)


 * Huh, you're right. He's only been here nine months. I've had so many annoying run-ins with him (such as him repeatedly reverting corrections I've made to acknowledged third-party errors, claiming it's ridiculous to expect him not to because it's not his responsibility to correct them) that I'd truly thought he'd been around for years. I've tried being nice, but have found that he doesn't respond in kind, so I'm in no mood to be more than barely civil to him. Well, if he's only been around that long, I've probably annoyed him just as much. kwami (talk) 01:34, 22 April 2008 (UTC)
 * Kwami, if we're going to discuss people rather than issues (which I think is a bad idea, but it's already being done here), you're one of the rudest people I've had to deal with in my entire life. You would benefit greatly from internalising WP:NPA.


 * I've never reverted a correction you've made. I have reverted erroneous edits you've made. You seem incapable of grasping the idea that reverting someone's erroneous edit doesn't magically give the reverter a duty to work further on the article in question. That's a fairly basic concept. Timeineurope (talk) 19:20, 22 April 2008 (UTC)


 * TE, you have reverted what you admitted were corrections that I've made to pronunciations because you didn't like the formatting or, in several cases, the way I used a comma. That's simply perverse in an editor.
 * I've done no such thing. Stop lying. Timeineurope (talk) 16:19, 23 April 2008 (UTC)
 * When Æµ§œš¹ asked me to show you how not to edit war, s/he was mediating, diplomatically trying to shame me into behaving better. Such attitudes are to be commended.
 * I can't read Aeusoes1's mind; he did make it sound like I was the only one 'edit warring'. Timeineurope (talk) 16:19, 23 April 2008 (UTC)
 * As for German pronunciations in English, that's true for every foreign name. Our readers aren't stupid, that we need to point it out.
 * If it's true, then why not point it out? Timeineurope (talk) 16:19, 23 April 2008 (UTC)
 * We don't say that "Paris" is pronounced in English, or that Beijing is . Of course, if you can convince the folks here at MoS that we should do this, fine. But I object to you insisting on changing the organization of the encyclopedia in the face of others' objections. kwami (talk) 20:24, 22 April 2008 (UTC)
 * 'Changing the organization of the encyclopedia'? Hyperbole much? This is about the inclusion or non-inclusion of one word in one article. Timeineurope (talk) 16:19, 23 April 2008 (UTC)


 * It sounds like a judgment call to me. Since you do hear something close to the French pronunciation of "Paris" in English-language film and TV from time to time, I wouldn't personally have any objection to putting that pronunciation in brackets.  (Along the lines of the guidance on this page: "If the pronunciation in a specific dialect is desired, square brackets may be used".) - Dan Dank55 (talk)(mistakes) 20:38, 22 April 2008 (UTC)
 * I do believe the Frenchier pronunciation of Paris that English speakers make ( right?) would still constitute code switching. — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi]  20:59, 22 April 2008 (UTC)
 * I'm not saying that using a German pronunciation in an English sentence isn't code-switching, I'm saying that the word 'also' needs to be added to Johann Pachelbel lest people will think that his name is never pronounced with [x] in English sentences. When we say that something is pronounced in a particular way 'in English', we are saying it's pronounced that way in English sentences, not necessarily using English phonology. Timeineurope (talk) 16:19, 23 April 2008 (UTC)
 * You're conflating two different things: Pronouncing Pachelbel with an [x], which can be done by people who don't know German, just as with Bach, and pronouncing Pachelbel in German. Pronouncing Pachelbel in German is not pronouncing it in English, which is what your edit claimed. Also, there are always going to be idiosyncratic pronunciations. Are we to add "plus many idiosyncratic pronunciations" to every single word we give a pronunciation for? If you can find any evidence that people anglicize Pachelbel while retaining the [x], then that should be stated explicitly, not mere implied with an "also". kwami (talk) 17:52, 23 April 2008 (UTC)
 * Actually, I find that pronouncing it in German in an English sentence is pronouncing it in English for these purposes. Readers will want to know how to pronounce Pachelbel when speaking English, and the current version of Johann Pachelbel will be interpreted as saying they must pronounce it with /k/ when speaking English. We all agree that that's not true. How about a compromise where 'also' is replaced with 'often'? Timeineurope (talk) 21:34, 27 April 2008 (UTC)
 * That's not a compromise. You're still claiming that English speakers speak German while speaking English, which is seldom the case. You might as well claim that "chocolate" is "often" pronounced with a final, or that the x in "Mexico" is pronounced . If you want to add an English pronunciation with an [x], fine. But I object to the claim that German is English, or even "often" English. kwami (talk) 00:47, 28 April 2008 (UTC)
 * You misunderstand. 'Often' would apply to the English-only pronunciations. Now try again. Timeineurope (talk) 01:15, 28 April 2008 (UTC)
 * Since you're not responding, I've implemented my compromise suggestion. Timeineurope (talk) 17:22, 29 April 2008 (UTC)


 * I'm not responding because you're arrogant and rude. Now you try again. kwami (talk) 17:37, 29 April 2008 (UTC)
 * You need to actually voice an opinion on the suggestion I made, rather than the suggestion you thought I made. Timeineurope (talk) 17:57, 29 April 2008 (UTC)
 * There are enough people who give Paris this pseudo-French pronunciation that it's universally recognized, so I wouldn't object to putting it in. It's similar to Bach, which a lot of people give the pseudo-German pronunciation . But Dan, it's hard to argue that the Parisian, or the French or German , occurs in English. Similarly, I wouldn't want to say English has tone, just because some people who know Chinese pronounce Chinese names with tone when they're speaking English. kwami (talk) 21:08, 22 April 2008 (UTC)
 * There's a danger when Anglicising pronunciations that not enough notice is taken of differences in how certain non-native sounds are Anglicised. Pachelbel is a case in point: there's a claim made above that the English pronunciation "should be", but in British English it is much more likely to be  (see John Wells's bolg, 16 Sept 2006).  This is quite common when the original has : many British English speakers (including me) have  for  and don't really perceive a significant difference between this vowel and the French or German one.--JHJ (talk) 08:15, 23 April 2008 (UTC)

←This is just the kind of thing you want the dictionaries to tell you, and just the kind of thing they usually don't, damn them. They either do or don't give the foreign pronunciation; either way, that doesn't answer the question of whether English-speakers tend to hear it, and if so, what they hear. My WP:OR is that is not generally heard in the U.S., even from the people who tried to teach me French in high school. If that's true, then I think it would be a misrepresentation and disservice to our readers to give a pronunciation including that sound. If they want to know how the French pronounce it, fr.wikipedia.org and other resources are a click away. But surely there's some kind of reference that can tell us about the prevalance of various pronunciations among Americans, the British, etc? - Dan Dank55 (talk)(mistakes) 04:15, 23 April 2008 (UTC)


 * For foreign proper names, I would advocate giving both the "local" pronunciation and the most common Englicised one. For example Paris (, locally ). &minus;Woodstone (talk) 08:16, 23 April 2008 (UTC)


 * Saying "locally" does deal with my objection about misinformation, as long as people understand what that means. Will the typical reader follow that?  Also: you're just saying that it's okay, right?  We're not going to make that a style guidelines recommendation for every place name? - Dan Dank55 (talk)(mistakes) 15:12, 23 April 2008 (UTC)


 * I think we need to first distinguish English from non-English pronunciations, then general vs. local pronunciations. Dan, I think that if the local pronunciation is truly different than the general pronunciation, by more than the difference in dialect would allow you to predict (different vowel reduction or number of syllables, say), then yes, I think we should make it MoS policy to list both. kwami (talk) 17:52, 23 April 2008 (UTC)


 * Quick note, I'm very busy with other WP stuff and it will probably be a week before I can return to this discussion. I'm still interested. - Dan Dank55 (talk)(mistakes) 19:22, 23 April 2008 (UTC)
 * I don't see a way to briefly indicate that using the foreign pronunciation while speaking in English isn't stigmatized. I'm also having difficulty believing that many people would even believe that to be the case.  If anything, it would be the opposite; giving the English pronunciation informs the reader that a pronunciation with no foriegn sounds isn't stigmatized when speaking English.  — Æµ§œš¹ <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA"> [aɪm ˈfɻɛ̃ⁿdˡi]  22:05, 27 April 2008 (UTC)

Audio pronunciation task force
I've created a task force in WikiProject Spoken Wikipedia, the Pronunciation task force. It's my attempt to organize and give direction on how to contribute audio pronunciations of difficult terms to Wikipedia. I've created two categories, Category:Requests for audio pronunciation for articles that need pronunciations, and Category:Articles including recorded pronunciations, each subcategorized by language. I'd like to add a section to this guideline with a bit of info and a link, if everyone agrees. Also, I'd love for anyone who is interested to join the task force, particularly native speakers of other languages. Any feedback is appreciated. Thanks! Dcoetzee 00:32, 27 May 2008 (UTC)
 * It might be useful to consider coordinating efforts with Wiktionary, they've already got a bunch of audio clips and it would also be helpful if similar icons/templates/etc. were used. It would be nice if we strove for representation of a wide variety of accents rather than the old RP vs GenAm. J IM ptalk·cont 03:43, 27 May 2008 (UTC)
 * Yup, I'm familiar with the Wiktionary efforts and I am borrowing some of their organization. I agree that it would be nice to represent a diversity of dialects, where applicable, although there would have to be some discussion around where these would go in the article - it would clog up the lead to put them all in there. Perhaps there should be some kind of "alternate pronunciations box" off to the side or in the infobox? Dcoetzee 22:33, 27 May 2008 (UTC)
 * I've just checked and Wikipedia's audio is very similar to Wiktionary's. The icon is part of the audio file ... or something ... so that's identical.  We might not be able to collect enough audio clips to start worring about clogging up the lead but if we do, perhaps we could put them in a footnote.  One approach might be to rework  to handle multiple clips giving something like "audio (Can, US) (help·info) ". J IM ptalk·cont 00:52, 28 May 2008 (UTC)
 * It looks like the icon is put in by the span class unicode audiolink. J IM ptalk·cont 01:05, 28 May 2008 (UTC)

Can't We Come To A Compromise?
Surely there is room for two simultaneous approaches. For the linguists, non-English speakers and perfectionists, by all means have IPA as the standard pronunciation guide. For the vast majority of English speaking casual users (substitute French, German, etc. for wikis in other languages) who just want to have a reasonable approximation of the correct pronunciation, include a (much) simpler pronunciation guide. When I look up "cruithne" I don't care if I am pronouncing it perfectly, or whether I pronounce it like I was from New Zealand or South Africa. I just want to know if it is pronounced as krooth'nee or kree'nuh or something totally different. That way if I use it in speech with someone who is familiar with the word, they will understand what I am talking about. 216.36.132.66 (talk) 14:54, 5 June 2008 (UTC)


 * This is already specifically allowed in the MoS, in addition to the IPA. I once tried instituting pronunciation respelling as a general policy, but it was roundly rejected. However, if you're interested, we came up with a standardized format, now at Help:Pronunciation respelling key. —kwami (talk) 23:09, 5 June 2008 (UTC)