Wikipedia talk:Manual of Style/Pronunciation/Archive 4

Suggestion for practical IPA usage in Wikipedia

• IPA stands for International Phonetic Alphabet. It's what you see in introduction pages of dictionaries for many languages, most certainly English dictionaries. Which readers – even as one had put it, in Africa – of the English Wikipedia might not have used a dictionary of this acquired language? Perhaps some users do not have a proper Unicode character set yet, but a complete UTF Arial font exists already and in informatics new things tend to get widespread in a very short time. The more exotic one's native alphabet, the sooner one will install such font. We are building Wikipedia also for the future: it would be unwise to introduce some unfamiliar phonetical references set for which there will be no need any more after one or two years.
• On the other hand, the link IPA redirects to International Phonetic Alphabet showing a decent linguistic description. As the cümbüş discussion here above illustrates, most people will start reading without even noticing the link to an English language approximation: it is hidden in what everyone sees as a disambiguity header while the underneath article looks as the relevant one. I might suggest putting that link in one of the first lines of the article proper, and let it stick out, so non-linguists would spot immediately their quick guide to a pronunciation – in fact that's what the article 'Wikipedia:Manual of Style (pronunciation)' does: one might put it much like that in the 'International Phonetic Alphabet' article to which IPA redirects. Though such guideline will only give a rough idea (inexact transliteration for some non-English sounds and again inexact interpretation of what the English is supposed to sound like), such will suffice for most users.
  

Better still, one should redirect 'IPA' (the article linked near any [<IPA characters>] in some normal article) towards the article 'IPA chart for English'. (I made a minor modification of that page's header line, making also such redirection appropriate and I assume acceptable for everyone.)
Advantage: impatient non-specialists get their thing immediately, anyone else (probably a minority of the occasions anyone actually clicks on 'IPA' near an IPA-transcription, because specialists do not often need the information) will know well enough which next link to follow.
Within any chart, an IPA character could link to a sound file so (most) users would simply and accurately hear the proper sound.

Note: Wikipedia standards should not allow linking to these sound files directly from an IPA representation on a normal page: users would correct some character but might forget to match the sound file, which would after a while cause a lot of confusion. The linking of sound files from charts should be protected against vandalism and against changing to what a person might think some looked-up word (in his particular dialect or language) and 'thus' an IPA character must sound like.

SomeHuman 2006-06-25 21:56 (UTC)

I'm not sure I quite understand this all, but you're saying if I want to know the pronunciation of something, I have to click on a link, scroll down and match up each character with a sound. Most people aren't going to want to take the time to do that, I think. I have a short attention span and I like to see the pronunciation right there before me on the page in a language I understand. --Liface 20:13, 25 June 2006 (UTC)

That's why it is allowed for some cases to write in a text e.g. "stress on the second syllable" or "rhymes with...". For rendering of the sound of a complete word, nothing simple will ever be able to give a good idea for the many words one may need help for (except a separate sound file but that's hard to create for several technical and practical reasons). Since you spent some time in discussing the matter, I assume you needed such help rather often. Once you start looking up IPA characters in a – as I suggested – directly linked easy-to-use table, you will very quickly remember some of the IPA and no longer need to look them up (as my personal experience with dictionary tables has proven – I'm just a layman too). -- SomeHuman 2006-06-25 20:34 (UTC)

The IPA chart for English is only suitable for looking up the pronunciation of English words. In many cases the words for which a pronunciation is given would be a non-English word. In that case it would be a mistake to use that table and recourse should be taken to the full IPA table. There are many sounds and symbols that do not occur in English. Apart from that, it is customary to use rather wide phonemic symbols for one specific language (c.q. English) that lead to mistaken phonetic value if applied to another language. For example, assuming that the symbol [r] in an Italian word is the same as [r] in English is quite wrong. −Woodstone 21:16, 25 June 2006 (UTC)

That's why I edited the initial line of IPA chart for English (if one does not revert, see its talk page under 'Linguistically detailed'): A link to a more proper but for many users too refined and illegible article must stand out but not be the first article these users (and most clicks on 'IPA' will come from them) arrive at to figure it all out. Linguists must not expect an average user to be interested in the particular difference between Italian and English pronunciation of some character, the latter user may often not even be able to hear, let alone to produce the correct sound of a non-English language; the speakers of such language will not click on 'IPA' because they know what 'their' word sounds like. My suggestion is a practical solution, a perfect and general one can not exist (not even with sound). Note: I'm not saying the 'IPA chart for English' might not be improved (to establish such or do something about it, is out of my league) -- SomeHuman 2006-06-25 21:59 (UTC)

SomeHuman writes: "[IPA is] what you see in introduction pages of dictionaries for many languages, most certainly English dictionaries."
I'm puzzled by that last phrase. You appear to be unaware that few - perhaps none - of the major American general-use dictionaries employ IPA. Does this make any difference to your thinking?--Chris 00:42, 26 June 2006 (UTC)

I was aware of this, but can't see how it might change the suggestion: it merely allows Americans to tackle a problem a lot of people (also outside the UK) encountered by the same easy-to-use IPA table. You would have a point in case the American dictionaries would commonly use an alternative standard, but to my knowledge they don't. Perhaps the need was not felt as much in the US because there are not the many old dialects of people that for generations stay in a same locality, by not having a number of old colonies, by not living in the neighbo(u)rhood of countries with dozens of quite different languages... and for those reasons perhaps people in the US may not even often want to click on the link under 'IPA'. But if they do, as speakers of (American) English, the 'IPA chart for English' is going to be a lot more helpful than the 'International Phonetic Alphabet' article they have been seeing so far. -- SomeHuman 2006-06-26 01:41 (UTC)

I agree completely, as I said in the semi-ranting continuation to my comment in the next section. I have no objection to bringing people up to speed on the IPA. However, I believe that in the short term (and this is likely to be one or two decades, not one or two years) a dual system will be optimally helpful for the user community as a whole - and, frankly, respectful to it. Occasionally I get driven into intemperate commentary by what I perceive as the doctrinaire position of some IPA-only proponents.--Chris 01:52, 26 June 2006 (UTC)

Makes sense, but see also my 2nd reply in your section 'Fuzzy thinking and unintentional cruelty' hereunder (and linguists, pay extra attention to the words 'some common sense' therein, please). -- SomeHuman 2006-06-26 02:51 (UTC)

I object strenuously to mangling encyclopedia articles and redirects for project-management reasons. IPA is the IPA, and we should not be redirecting that to IPA chart for English (which is clearly not the actual IPA, and linking like so violates all kinds of editorial guidelines) for infrastructural reasons. If such a need is truly great, there ought to be a Wikipedia: namespace page to contain a pronunciation guide. We do not co-opt Encyclopedia for a description of Wikipedia, and neither should IPA or IPA chart for English be chain-ganged into being anything less than the absolute best informational encyclopedia articles we can make them. — Saxifrage ✎ 05:34, 26 June 2006 (UTC)

The article 'IPA chart for English' starts with this line (and that rendering in bold):
.   Of the linguistically detailed International Phonetic Alphabet (IPA), this page is a concise version for English sounds.
I can't imagine anyone assuming the chart underneath would be the absolute best informational encyclopedia article on International Phonetic Alphabet (IPA) – it is pretty much like a disambiguity page – thus any reader who wants that standard of quality, will simply click the very first link on that line. The important part is that wherever a word's pronunciation is spelled by IPA characters in an article, the acronym 'IPA' must not send people to a page they simply can not use. Wikipedia is not the Standard Reference Work of Linguistics, it is an encyclopedia for everyone and its quality improves by making it accessible. Please see "Ignore all rules". -- SomeHuman 2006-06-26 06:33 (UTC)

(Invoking Ignore All Rules in an argument is, I believe, one of the ways of losing an argument by virtue of the very nature of that meta-rule. As a courtesy, I'll ignore the invokation beyond this.)

Make it accessible by doing it right, then. Abusing encyclopedia articles that are about a subject (IPA chart for English is decidedly not a disambig page or other infrastructural page: it's an article) by adulterating them with material that is Wikipedia-specific is not done. See Wikipedia:Avoid self-references for a grounding in why; I believe that spindling, folding, bending, and otherwise mangling an article for Wikipedia-project purposes (as opposed to quality of article purposes) goes against the same spirit.

Just to put this in context, I grant you that the change you made to IPA chart for English is small. I'm arguing the principle that it shouldn't be done at all so that any larger-scale co-option of articles can be nipped in the bud. We have the project's manual of style at Wikipedia:Manual of Style, not at Manual of style. Neither should we use IPA nor IPA chart for English as a replacement for Wikipedia:Pronunciation guide. We must keep our article space clean of project contamination. — Saxifrage ✎ 21:28, 26 June 2006 (UTC)

Fuzzy thinking and unintentional cruelty

I tend to agree that the IPA link should access the IPA Chart for English. Proponents for the IPA-only solution have long argued that a link to the general IPA page was all a novice needed to orient themselves. The fact they would make so patently ridiculous a claim (see cümbüş above) suggests that they haven't thought through their arguments properly. You at least have. If the average English speaker was as linguistically sophisticated as the average Dutch speaker appears to be, an IPA-only solution would be desirable. As it is, insisting on IPA only seems an act of unintentional cruelty. Those who already know IPA well (that is, well enough to interpret the various conflicting transcriptions in use in WP), or who are able to quickly master it through the IPA Chart for English or similar are least likely to actually need help on pronunciations. They will already know how to pronounce Ayn Rand, Bruggen, Houston Street, and so on. Those who need the help are precisely those least likely to know or be able to quickly learn IPA. In many cases, a dual solution (IPA + a intuitive respelling key) will be the most practical.--Chris 01:04, 26 June 2006 (UTC)

(See also my reply on your edit at 00:42 today in the chapter with my suggestion, please). I'm not implying people have to learn IPA: the IPA is a shorthand that does not take too much room and, unlike spelling-it-out (which is very often misleading for non-native speakers of English), it can be easily disregarded if not to the reader's field of interest: Everyone who ever saw this shorthand will immediately recognize it as a pronunciation indicator. The simple chart allows a fast look-up of whatever part (usually just one or two IPA characters) one can't immediately figure out by the normally written word. Believe me, it's really not going to be unbearably cruel at all – I know because I still use it myself, since I do not master IPA, it just helps me out. -- SomeHuman 2006-06-26 02:09 (UTC)

We all seem close to agreeing on something here. Is it acceptable to say that both a "rhymes with" or "sound-out" pronunciation and IPA pronunciation should be sufficient for most hard-to-pronounce article subjects? --Liface 02:12, 26 June 2006 (UTC)

The "rhymes with" or "stress on nth syllable" is not quite "sound-out": the latter would often require a too long series of approximations and force paying attention to the pronunciation. More often just IPA will do fine, but one must not expect every writer of an article to master IPA and (even if one has some IPA knowledge) it will be easier just to sound-it-out. Helpful linguists might later on insert the IPA symbols and, with some common sense decide that the sound-out is not really needed and omit it for easier reading of the article's real topic. -- SomeHuman 2006-06-26 02:32 (UTC)

The statement that the people who know IPA already know how to pronounce everything is wrong. I know IPA but I'm totally unfamiliar with, for example, German orthography, so when I look up Wittgenstein the IPA is the only thing that tells me how to pronounce it correctly. Something like "VIT-gun-stein" would be useless to me. I strongly doubt I'm the only person in this position. You can't divide everyone into "people who know all about language" and "people who are totally helpless". —Keenan Pepper 15:49, 26 June 2006 (UTC)

I agree of course; it's just a tendency. The Keenan Peppers of this world are more likely to know how to pronounce "Wittgenstein" than some grade 8 student -- if only from the Monty Python Philosophers' Sketch :-). Vitˈ-gən-styne might be of some use (note anglicised pronunciation!), though [aɪ] is notoriously hard to represent in a intuitively obvious manner. The IPA version tends to get pronounced vitˈ-gən-stainby the naive, and if there's an audio file whoever does it will probably insist on a full Austrian pronunciation - so there's no way to win--Chris 16:25, 26 June 2006 (UTC)

Gentlemen, please, as we are trying to understand each others views, we should not focus on our own. Of course a 'sound-out' can not possibly be used for all non-English sounds (and there are many in the world's languages). The idea is to have IPA at all times a pronounciation indication is required, but very few authors will be able to produce a correct IPA themselves: for a while a 'sound-out' may be the only thing available. When someone (like Keenan) spots it later on, he might create the IPA characters string. Should then a 'sound-out' remain? Is such properly standardized? I'm not very familiar with it but do not think so: I saw the 'pro-nun-see-AY-shun' and wonder how 'pronoun' would look as 'sound-out': the first syllable of 'pronunciation' should not sound identical to the one of 'pronoun'. How could one realize that '-see-' has a short vowel? The '-AY-', not preceeded by a consonant, to me looks like Aye-aye thus I would expect it to sound like the word 'eye'. That's just one word 'pronunciation' and I would be left uncertain or dead wrong three times. I'm afraid the easily read 'sound-out' will refrain people to look at the IPA chart and they might then have been better off without any pronunciation indication at all. Some words might be easily represented by 'sound-out' without much risk of anyone who reads English to misinterpret – is there some source (American dictionary?) that recognizes and tackles the problem? In other words, instead of arguing, let's find out whether it can work and which restrictions there would be. -- SomeHuman 2006-06-26 22:59 (UTC)

SomeHuman, I appreciate your constructive approach. I will try to write up a longer response tomorrow.--Chris 06:22, 27 June 2006 (UTC)

As the article Pronunciation respelling for English shows, there is no standard. Practically every dictionary employs its own specific way of respelling. It is not possible to define a clear set of respellings, that are never confusing in any context. That is the major reason IPA is preferable. −Woodstone 21:28, 26 June 2006 (UTC)

Unfortunately, different dictionaries' interpretations of IPA vary almost as much the traditional respelling keys. For instance, the Canadian Oxford and the Shorter Oxford use [a] for the vowel in cat, while the Oxford ESL Dictionary uses [æ]. The vowel in pipe is [əi] in the Oxford Canadian, [ʌɪ] in the Shorter Oxford, and [aɪ] in the Oxford ESL. And that's not even touching on the rhotic/non-rhotic distinction. IPA, though invaluable for many purposes, is no panacea.--Chris 06:17, 27 June 2006 (UTC)

The story so far

This discussion has meandered under several headings. I will attempt a summary. The options appear to be:

• Sound files
  • Obviously useful for giving "what the word sounds like" if the speaker is clear and a native speaker of the relevant language/dialect/accent
  • Impractical as a sole solution, in any context
  • Does not articulation of the vocal tract explicit (which is relevant in articles discussing phonetics in detail).
• IPA transcription [prɘˈnʌnsiˈeɪʃɘn]
  • Essential for non-English words, and when discussing specific English accents, dialects, phonetics, speech impediments, etc; makes articulation of the vocal tract clearly explicit.
  • Those unfamiliar with IPA may have difficulty interpreting symbols.
    • For pronunciation of English words this could be reduced by linking to an English-specific IPA guide rather than the general IPA page.
  • For pronunciation of specific English words, will inevitably be specific to a particular accent. Those who speak with a different accent will be obliged to extrapolate accordingly.
  • Needs IPA template for IE readability
• Ad-hoc desciptive workarounds "rhymes with nation"; "stress on second and fourth syllables"; "the C has an S sound"
  • Easy to understand
  • Only possible for some cases
  • May be verbose
  • "rhymes with" may be accent-specific
• Ad-hoc respelling pruh-NUN-see-AY-shun
  • Sometimes easy to understand for native speakers of English
  • sometimes ambiguous, through poor choice of representation, or almost inevitably for certain sounds (th for [ð]/[θ]; oo for [u]/[ʊ]) unless combined with descriptive workarounds
  • sometimes rely on assumptions about accent (especially when attempting to approximate foreign words)
  • often difficult for nonnative speakers of English
  • Possibly as a stopgap on a particular article, this is better than nothing at all, provided the accent of the editor attempting the transcription is made clear; can be replaced by a better method by a competent subsequent editor.
• Standardised respelling prɘ'nŭn.sē'ā.SHɘn
  • Need to interpret via a key
  • Key may be more intuitive than IPA for English-speakers, especially those familiar with American dictionaries
  • Can be applied to a variety of accents; more symbols will allow more mergers and splits to be accommodated, though at the risk of misrepresentations in a transcription by somone whose accent has the merger.
  • Would be specific to Wikipedia as no external standard exists (though I guess we could plump for any one of the many in use)
  • Only usable for English
  • Depending on symbols chosen, may need a template similar to IPA for IE readibility

Questions for the floor:

1. Which of these substantive points are in dispute?
2. Any points missing?
3. Based on these points, what option(s) should we pursue?

jnestorius^(talk) 14:02, 30 June 2006 (UTC)

• Sound files

Symbolic notation can make articulation of the vocal tract explicit, whereas even a clear speaker's speech can be difficult or impossible to analyze.

• IPA transcription [prɘˈnʌnsiˈeɪʃɘn]

IPA also makes articulation of the vocal tract clearly explicit.

• • Those unfamiliar with IPA may have difficulty interpreting symbols

This is equally true for all systems. I would argue that IPA is more intuitive than others (see below).

• • For pronunciation of specific English words, will inevitably be specific to a particular accent. Those who speak with a different accent will be obliged to extrapolate accordingly.

I don't believe this is so. Although we haven't developed a specific convention in Wikipedia, a 'broad' IPA transcription can be used for English words which uses only the more basic Latin symbols, and allows, e.g., an American, British or Scottish reader to pronounce the letter /r/ in their own way. As a convention, square brackets [...] could be used to indicate phonetically precise IPA for foreign words, and slashes /.../ could denote phonemic English IPA. Where necessary, phonetic notation in brackets could be used to illustrate precise details, e.g. comparing different English accents.

• • Needs IPA template for IE readability

Is this true for the basic English phonemes?

This is true of the old-fashioned dictionary respelling systems too. The following symbols break in MSIE 6 on a vanilla Windows XP system: o͝o, o͞o, ʉr, s​͡h, th̸, K​͡H. Some also require HTML: <i>th</i>, <small>K​͡H</small>.

• Standardised respelling prɘ'nŭn.sē'ā.SHɘn

This is not standardized: see the inventory at Pronunciation respelling for English

• • Need to interpret via a key

IPA interpretation can be similarly assisted with a simple key for basic English phonemes

The IPA vowel and consonant charts are also logically arranged according to place of articulation, to help readers interpret the full range of international sounds.

• • Key may be more intuitive than IPA for English-speakers, especially those familiar with American dictionaries

It may feel more familiar to some readers, but it's less intuitive than IPA. For one thing, the many systems have subtle variations so a key is absolutely necessary, for another, IPA characters have logic:

• IPA uses plain English Latin letters for many sounds: /a, e, i, o, u, b, d, f, g, h, k, l, m, n, p, r, s, t, ts, v, w, z/
• IPA uses letters representing sounds as in other languages, whose use may be familiar to many readers, even if they aren't fluent in the other language: /θ/ from Greek, /æ, ð/ from Old English, /j, x/ from central European languages, /y/ from French.
• The more subtle phonetic variations, when needed, are variant forms of the related Latin letters, assisting learning: /a, ɑ, ɒ, ʌ, æ/, /e, ɛ, ɜ, ə, ɚ/, /i, ɪ/, /o, ɔ/, /u, ʊ/, /n, ŋ/, /r, ɹ/, /s, ʃ/, /w, ʍ/, /z, ʒ/.

• • Can be applied to a variety of accents; more symbols will allow more mergers and splits to be accommodated, though at the risk of misrepresentations in a transcription by somone whose accent has the merger.

This could also true of phonemic IPA for English, denoted by slashes /.../, see above.

—Michael Z. 2006-06-30 15:23 Z

I added to the summary based on your response; some of your points I thought were already addressed. To clarify: by "Standardised respelling" I mean standardised within Wikipedia (with a key on a Wikipedia: or Help: page) as opposed to each editor rolling their own for each word they give a guide for. I don't think there is any dispute that IPA is needed for non-English words, and for non-native-English-speakers, and where phonetic detail is needed.

As regards using IPA for accent-neutral English phonemes: is there any standard for this? Can you give some sample transcriptions? It seems to verge on what I have called "Standardised respelling" except that the symbols chosen would happen to be IPA symbols. Whether this is better for using "standard" symbols or worse for stretching the definition of "phonemic transcription" is certainly worth debating. jnestorius^(talk) 16:43, 30 June 2006 (UTC)

I agree with Michael Z's first note: "Sound files: Ideal if the speaker is clear and a native speaker of the relevant language/dialect/accent" is only true for that language's native listeners, others seem to hear what they expect or remain quite deaf to some differences in sound (e.g. many Germans, Dutch and French have a problem correctly hearing the sounds represented in normal English writing by 'th'. (One might spot their nationality by their way of attempting to pronounce the English sounds). The phonetic symbols are then more easy for them to differentiate. Native speakers of English have the same problem with some foreign sounds. -- SomeHuman 2006-06-30 16:45 (UTC)

Jnestorius, a standardized phonemic IPA for English would work essentially the same way as another spelling key, and a simple key would help unfamiliar readers (it could be simpler than the existing IPA chart for English). Examples exist (e.g. in any English dictionary that uses IPA, but I'm not familiar with one that is meant to be accent-neutral). Using IPA this way has the advantages that it is compatible with international IPA, and that a standard convention for indicating it exists (/.../ vs. [...]).

SomeHuman, IPA can convey information even to a native speaker which can be hand to divine when listening or even speaking. Articulation has subtleties that speakers don't think about in natural speech, but this issue is relevant to detailed phonetic analysis, and not to conveying the simple phonemic pronunciation of English words. —Michael Z. 2006-06-30 17:18 Z

An example of the Canadian Oxford Dictionary's IPA for Canadian English is transcribed at Talk:Canadian English/Archive 1#Vocalic differentiation of Canadian English from American English—it has a few generalizations specific to Canadian English, for example the commonly-used pronunciation of the letter a [æ] is transcribed as /a/, and adds some French vowels used by Canadian anglophones. For Wikipedia use, it would use some similar conventions, but somewhat more generalized: e.g. /a/, /æ/, and /e/ could suffice to cover most of the subtle variations of these vowels; /r/ would be suitable for the dialectic American [ɹ], British non-rhotic [ɑː], Scottish rolled [r]. —Michael Z. 2006-06-30 17:37 Z