Help talk:IPA

Proposal

This is a proposed site for a future chart to quickly explain the values of the symbols of the IPA. Once it's up and running, {{IPA2}} can be redirected here.

People have been asking for a simple key to the IPA for quite a while. It's not an easy request, but I propose that we make a list with the symbols grouped graphically, so that e.g. all symbols derived from a are together, because newbies won't know how else to look them up. Each symbol should be linked to its own article, since a lot of the time a simple chart won't be of much help. Then there should be a brief "good-enough" description ("[t] with the tongue curled back" etc.), avoiding technical terms such as "back vowel" or "labial" or "fricative"; and examples, if there are any, from languages native English speakers might be familiar with, such as RP, French, German, etc. I think that if there isn't any reasonably well known language that has the phone in question, we should simply leave that section blank, and let the reader navigate to the page in question. How does that sound? kwami 03:56, 18 October 2007 (UTC)

Format

Excellent start, I'll be glad to help building. How about reformatting like:

Symbol	Examples	Description
A
[a]	Spanish casa, French patte	For many English speakers, the first part of the ow sound in cow.
[ɐ]	RP but	(With English, this is normally written "ʌ")
B
[b]	English bat
[ɓ]	Swahili bwana	Implosive [b]

−Woodstone 07:55, 18 October 2007 (UTC)

Hi Woodstone. I didn't see this until after I pretty much finished the chart (though I still need quite a few examples), so I'll just give you my reaction:

• Pro: Your format looks much nicer than mine. It's more solid and uniform, whereas mine is broken up and uneven.
• Con: My format enables an index, so the reader can navigate easily, and it allows more variation in column width, which means the columns aren't wide unless they need to be to fit their contents.

I don't mind one way or the other, so reformat it if you like. kwami 09:04, 18 October 2007 (UTC)

P.S. If you can link the symbols to their individual articles (but keep them formatted as they are in the IPA article, without making them bold or underlined), then I think we are pretty much done. Get some feedback, and maybe we can link up to the template. kwami —Preceding comment was added at 09:07, 18 October 2007 (UTC)

I put in some links in the example above. I do not see underlines. I assume that is triggered by the IPA style in the table header. Do you see underlines? −Woodstone 12:57, 18 October 2007 (UTC)

No underlines. Evidently that's a property of the IPA class. kwami 18:49, 18 October 2007 (UTC)

I doublechecked: the stylesheet mediawiki:common.css has:

/* Remove underline from IPA links */
.IPA a:link, .IPA a:visited {
  text-decoration: none;

−Woodstone 19:29, 18 October 2007 (UTC)

Organisation

The chart is justly organised by resemblance of the symbols to latin letters. How do we deal with doubtful cases. I saw for example /ʌ/ duplicated under A and V. For this case clearly the intention of IPA is similarity to A. How do we deal with this? The most logical place only? Duplicate entry? Or a line at V like "/ʌ/ see under A".? −Woodstone 15:49, 18 October 2007 (UTC)

I think the last of those is the best.--JHJ 17:21, 18 October 2007 (UTC)

The table's so short that I don't see a problem with duplicating. I put [ʒ] under both G (its historical source) and Z (its graphic source), for example; under G the example is beige, while under Z the example is azure. I'm hoping those will act as mnemonics no matter which one people look up. kwami 18:35, 18 October 2007 (UTC)

If we duplicate entries, they will grow apart in future edits and lead to confusion. It would be better to refer to the section where they are described, as I have done for most of them already.−Woodstone 19:22, 18 October 2007 (UTC)

Some queries

I'm not at all convinced by the description of the lateral fricatives mentioning [ʃ] and [ʒ]. English speakers often approximate the Welsh sound with some English voiceless consonant plus /l/, but they don't choose a sibilant in my experience. (And it doesn't sound anything like [ʃ] to me.) Is there a reason for this choice?--JHJ 17:21, 18 October 2007 (UTC)

It often seems to have such a quality to me, and when I describe it this way some people seem to have an easier time getting it. What else would we say?

That seems like OR to me. We need to be very careful about basing such descriptions on our own perceptions; as I said, my perception is different: it (in the Welsh I've heard spoken, not the Wikipedia soundfile) sounds more like [θ] than [ʃ] to me. I'd just describe it as the Welsh sound.--JHJ 18:56, 18 October 2007 (UTC)

But most people aren't familiar with Welsh. Also, it's used for a great many languages around the world, and in those I'm familiar with in Africa and America it's not sibilant to be sure, but not at all [θ]. Maybe we could list both; that might be good enough for someone who just wants an idea, but serve as a warning that there's more going on. kwami 19:07, 18 October 2007 (UTC)

Maybe (and they do share the key feature of being voiceless fricatives), but I'm still concerned about relying too much on our perceptions.--JHJ 19:16, 18 October 2007 (UTC)

Maybe we can wait and see if it works for people, or if we get complaints that it only confuses or misleads them. kwami 20:12, 18 October 2007 (UTC)

Is [θ] missing, or did I just not see it?--JHJ 17:21, 18 October 2007 (UTC)

It was under 'other'. kwami 18:45, 18 October 2007 (UTC)

Front and back A

This is an old discussion, but there seems to be a reversal in the interpretation between back and front A between sources.

language	source	[ɑ]	[a]	[aː]
French	common practice	pâte	patte
German	IPA handbook		hatten	baten
	many dictionaries	hatten	baten
Dutch	IPA handbook	bad	zaad

Whereas the actual vowel sounds are like:

• patte = hatten = bad
• pâte = baten = zaad

We have never been able to solve this contradiction satisfactorily. How to proceed?

−Woodstone 18:21, 18 October 2007 (UTC)

As I understand it, a lot of French speakers have lost this distinction, and have [a] for both. For German, my understanding is that different forms of German do it differently. Given this, the German examples and the French one for [ɑ] may be confusing, and my suggestion is to remove them. We have enough other examples for both these vowels.--JHJ 18:20, 18 October 2007 (UTC)

The merger in French is supported by the article on French phonology. The article on German phonology, as far as I can tell, uses [a] for both the long and short vowels. I thought it used to comment on the variation, but I couldn't see that on a quick read-through.--JHJ 18:25, 18 October 2007 (UTC)

The German is confusing, but people still learn the French distinction. kwami 18:46, 18 October 2007 (UTC)

The Duden-Aussprachewörterbuch uses [a] and [aː], and that's the way they are actually pronounced in standard German. Timeineurope 18:51, 18 October 2007 (UTC)

The actual vowel sounds are actually:

hatten, patte: [a]

baten: [aː]:

bad (standard pronunciation): [ɑ]

pâte: [ɑː] or [a]

zaad (standard pronunciation): [aˑ]

Timeineurope 19:06, 18 October 2007 (UTC)

Prefer dictionary words

In quite a few cases names of places or persons are given as examples. This is not a good choice, firstly because they cannot be looked up in a dictionary, secondly because they are often pronounced irregularly. −Woodstone 18:06, 18 October 2007 (UTC)

The proper names given as examples can all be looked up in dictionaries, none of them are pronounced irregularly, and several of them are linked to Wikipedia articles with sound files. Timeineurope 18:39, 18 October 2007 (UTC)

Remember that a lot of people will not be familiar with these languages, so internationally known words and names, that non-speakers have been exposed to, are IMO preferable. kwami 18:48, 18 October 2007 (UTC)

However proper names are frequently mispronounced in other languages. Often just by reading them as if they were English words. −Woodstone. Who would ever guess that "paris" is [paʁi], not [pʰæɹɪs]? 19:19, 18 October 2007 (UTC)

A good reason to link to the article. If they don't cover the pronunciation adequately, the link will cause people to complain, it will be fixed, and Wikipedia will improve. kwami 20:10, 18 October 2007 (UTC)

[c]

The only language I'm aware of that has [c] is Hungarian, and that only in very formal speech, so let's just leave this one blank. The Malay and Turkish examples were wrong, and even if they were right, I fear adding new languages will create chaos where everyone will add their favorite language, and we'll get away from making this accessible. kwami 19:11, 18 October 2007 (UTC)

I don't see how. Surely the more languages added, the greater chance that the reader knows one of the languages. Timeineurope 19:15, 18 October 2007 (UTC)

I'm afraid that the list will unmanageably long. We'll end up with examples from Koaia under half the entries, with some editor getting upset that we're oppressing him by not allowing every allophone in his language to be given equal time. Another problem we can see with the next example: Illustrating with languages that are not sufficiently well described to be used accurately. (Okay, Turkish is well enough described, but shows the potential problems involved.) kwami

The Turkish example was from the IPA handbook. If a phone does not occur in any of the primary languages, examples from languages that have it should be allowed. I agree though not to add more languages than necessary to get coverage. −Woodstone 19:19, 18 October 2007 (UTC)

Two problems: First, the letter used is <ç>, not <c>, which represents [dʒ]. Second, <ç> is phonetically [tʃ], not [c]; it is represented by /c/ phonemically. This usage is commonly seen in Hindi as well, as noted in the description, but is imprecise. If we can use Turkish, we should also be able to use English church, or Hindi, which is also already one of our languages. kwami

20:07, 18 October 2007 (UTC)

The IPA handbook (edition 1999) shows for Turkish in the table [c], not [ç] and gives as example explicitly in the text that [c] occurs in /caɾ/ kâr 'profit'. −Woodstone 20:37, 18 October 2007 (UTC)

Oh, I thought you meant it was spelled <car>, in Turkish orthography! The work kâr has traditionally been analysed as /kʲ/, but I suppose it could be [c]. My recollection of Turkish, especially of Persian borrowings such as this, isn't good, but /k/ is [c] or something pretty close to it before front vowels, and perhaps in Persian borrowings such as kâr. If we use Turkish, we can get ɟ and ɯ as well, so maybe it's worth it. kwami 09:27, 19 October 2007 (UTC)

We happen to have a Turkish word with that sound in it in English, kebab, so that's helpful. I also added Zulu, because we can fill so many blanks with it. kwami 10:14, 19 October 2007 (UTC)

Linking example words

In my opinion we should not link the example words to wiki articles. Here they are words as words> Their meaning is irrelevant. −Woodstone 19:25, 18 October 2007 (UTC)

Often, however, the article gives information on the pronunciation, and in a way that might be more accessible to the novice than the phonetics articles. kwami 19:55, 18 October 2007 (UTC)

Suggestions to improve article

1) Hi, thanks to everyone who is working to build this article. One comment to improve the article from a person whose only knowledge about IPA is from looking at the pronunciation key in the Oxford English Dictionary: can the IPA symbols be made bigger? Some of them are very small and difficult to make out. Cheers, Jacklee 22:16, 18 October 2007 (UTC)

2) A more common alternative to the OGG sound files, such as MP3, would make this page more accessible.— Preceding unsigned comment added by 192.101.136.224 (talk • contribs)

Verification

I did a lot of mass editing on links, new symbols, spacing and symbol size. I hope I haven't broken anything. I may have selected a few wrong symbols, because they are difficult to read in the edit window. Please help verifying. −Woodstone 20:31, 19 October 2007 (UTC)

Some of the symbols are very rare, we don't have (and are not likely to have) any examples, nor can we explain them adequately. If we leave them in, someone will try to "fill the blanks" by adding confusing and very likely inaccurate data, so personally I think we should leave them out. They are the epiglottals, ʡ, ʜ, ʢ, velar ʟ, implosive ʛ, the ʘ, ǂ clicks, a couple sounds which arguably (per Ladefoged, who chaired the IPA for a time) do not actually exist (ɶ, ɧ), and a bunch of less common diacritics. If people get this far into it, they can use the main IPA article. kwami 22:02, 19 October 2007 (UTC)

Fully agree, those rare symbols were left over in the list I constructed by parsing the consonant tables in the main IPA article. I just added the most common ones, like [dʑ] (Chinese), [tɕ] (Chinese, Thai), [ts] (Russian) and a few small caps. By the way, the IPA handbook has for German achlaut the symbol [χ], not [x]. For [ɦ] it has Dutch <hoed> as example. Perhaps in the list of reference languages in the lead, we should make a distinction between the most familiar ones (that can be used in parallel), and other ones, that are only referenced if the primary ones do not contain a phone. That part may then be a longer list. I do not understand the rationale of the list as it is now. Perhaps they are the languages you (Kwami) know something about? −Woodstone 22:20, 19 October 2007 (UTC)

The problem with [x, χ], as with [a, ɑ], is that many languages are ambiguous. Different people, dialects, and social registers may have different values, and often their pronunciation is intermediate, so it becomes the call of the linguists to say which it is, and they disagree among each other. There are languages in the Americas and the Caucasus with unambiguous [χ], but then we're starting to get pretty obscure.

I'll probably be overruled eventually in the choice of languages, so I'll just give my thoughts here. I left out some languages I'm familiar with, because they're too obscure, or, like Japanese, don't add much of anything to the chart. (Most of the phones Japanese is supposed to add, [ɴ ɯ ɺ], don't really have those values.) On the other hand, a huge number of English speakers are familiar with French, German, and Spanish, and they fill in a lot of blanks, so they were my first tier. Then Mandarin, Hindi, Arabic, and Russian are big languages, and also fill in a lot, so they were second tier. (Hindi's good for the retroflexes, etc.) Swahili is good for the implosives, and then Turkish was suggested. These aren't going to help many people, and so I guess are third tier. Zulu is your best bet for clicks, but is pretty much window dressing, because who's familiar with Zulu but not the IPA? (Some, I'm sure, but not great numbers like with the other languages.) So even with me I'm seeing language creep, adding one, then another ... Luckily, there aren't all that many blanks left. kwami 01:18, 20 October 2007 (UTC)

Protection?

I'm wondering if we should protect this chart once it's linked to the IPA2 template. Since it will then be accessed from thousands of articles, it would be a great opportunity for mischief, and even random unsourced edits could confuse huge numbers of people.

There are a couple levels of protection: One, you have to be signed in to an account to edit, and two, you have to be an admin to edit. D'y'all think either of these would be appropriate? kwami 01:43, 20 October 2007 (UTC)

Let's try it with no protection first and see how that goes. If that doesn't work, we'll try semi-protection before full protection. Nothing should ever be protected preemptively. Timeineurope 06:32, 20 October 2007 (UTC)

You're probably right. Another thing we could add is sound files, or even link to the files on the UCLA site. The chart can still be cleaned up a bit, but the links are in place, so we're ready go live. kwami 06:52, 20 October 2007 (UTC)

Okay, we're now online. kwami 06:55, 20 October 2007 (UTC)

Do we really want to list long vowels separately?

Generally, vowels are only long in comparison to short vowels in the same language, if by "long" we exclude double vowels as in Japanese. That is, the distinction is relative rather than absolute. For example, can we really say that German Klee is long and, in comparison, French clé is short? Or that English sea is significantly longer than Spanish si? If not, then what we are indicating isn't absolute vowel length, but a phonemic distinction, and therefore doesn't belong in a phonetic chart. This is in contrast to aspiration, where milliseconds of VOT can be used as an absolute measure.

We have length listed in the diacritic section, and elsewhere a phonemic IPA chart for English, which should be sufficient for our readers. kwami 01:03, 21 October 2007 (UTC)

If we leave the long ones out, do we merge the long and short examples? Or do we select one of them? Next question will be: do we need the affricates ([ts] etc.): are there any cases where they contrast with the simple combination of the symbols? I'm ambivalent about both. Limiting each row to exactly one symbol has a lot of attraction. −Woodstone 09:41, 21 October 2007 (UTC)

For the vowels, I think just select one example. Those will mostly be from what we now mark long; in the case of [i] and [u] we could add a comment that in the case of English these are often written long.

For the affricates, I think [tʃ] and [dʒ] are worth keeping, because English speakers have a hard time hearing them as sequences, but I don't see any point to [ts] and [dz]. (People will wonder why we don't have [ks]!) Perhaps we should keep [tɕ] and [dʑ], though, since they're palatalized parallels of [tʃ] and [dʒ], and people will otherwise be confused by them. kwami 11:49, 21 October 2007 (UTC)

I fail completely to see how this would make the chart more useful, and I have no problem seeing how it would make the chart less useful. Why should we say that both Mann and Aachen are pronounced with [a] when there's both a phonetic and a phonemic difference between the [a] of Mann and the [aː] of Aachen?

There's clearly such a thing as absolute phonetic length. Vowels are not only long in comparison to short vowels in the same language, but also in comparison to short vowels in other languages. The difference in vowel length between the [e] of French clé and the [eː] of German Klee is quite striking.

An affricate [ts] and a sequence of [t] and [s] are phonetically different, just compare a typical pronunciation of the Norwegian word tsar with how the German word Zar is pronounced.

Timeineurope 12:52, 21 October 2007 (UTC)

I guess they're all keepers, then. Also, you might consider adding something about the difference between Norwegian tsar and German Zar to Affricates vs. stop-fricative sequences. That would be a nice addition. kwami 23:04, 21 October 2007 (UTC)

Sound samples

What do you think about adding sound samples, like in de:Liste der IPA-Zeichen? Korg (talk) 05:04, 22 October 2007 (UTC)

By all means do so. It was suggested above, but wasn't necessary before we went live. Also, I see that the De page has (Southern) German ch as uvular, which we've been fighting about here, so I changed our table to match. kwami 07:37, 22 October 2007 (UTC)

Done. kwami 16:30, 27 October 2007 (UTC)

Thank you! Korg (talk) 23:52, 13 November 2007 (UTC)

Template

The {{IPA}} template does nothing. The {{IPA2}} template looks like the {{IPA}} template for Wiktionary. Can these ( {{IPA}} and {{IPA2}} ) be switched for the sake of consistency (and ease and simplicity) between Wiktionary and Wikipedia? Also, there doesn't seem to be a {{SAMPA}} template here. I guess I'll just steal the SAMPA code from wiktionary.

Another issue is: Should pronunciation even go here in Wikipedia? Maybe we can just link to Wiktionary. Or, we might even be able to put information from Wiktionary per word into the Wiktionary link template. Erudecorp ? * 21:21, 24 October 2007 (UTC)

Read the documentation on the template pages. {{IPA}} forces an IPA-compatible font. Without it, Internet Explorer is not capable of displaying the IPA. Once Microsoft catches up with the rest of the world, we should consider doing as you suggest. kwami 22:28, 24 October 2007 (UTC)

As for your second point, most people assume this will only be used for the odd word people can't be expected to be able to pronounce. But the idea of a link from Wiktionary, so that the pronunciations can be keep consistent, is a good one. Does Wiktionary include personal and place names? kwami 00:36, 25 October 2007 (UTC)

Wikitionary
Pronunciation Template:EnPR, IPA: [nɒt], SAMPA: Template:SAMPA Audio (US)ⓘ
Etymology O.E. cnotta "intertwining of ropes, cords, etc.," from P.Gmc. *knuttan- (cf. Low Ger. knütte, Du. knot, O.H.G. knoto, Ger. Knoten. Fig. sense of "difficult problem" was in O.E. (cf. Gordian knot). Symbolic of the bond of wedlock, c.1225. As an ornament of dress, first attested 1400. Meaning "thickened part or protuberance on tissue of a plant" is from 1398. The nautical unit of measure (1633) is from the practice of attaching knotted string to the log line. The ship's speed can be measured by the number of knots that play out while the sand glass is running.

I used a collapsible table to make the example solution to the right. It is not a template, but could become one, I imagine. This could clear up and make room for proper leads. Note that it emulates proper Wiktionary layouts. Erudecorp ? * 06:17, 25 October 2007 (UTC)

The pronunciation is an essential part of a proper lead. Timeineurope 12:29, 25 October 2007 (UTC)

This wouldn't exclude that information. The pronunciation is already on Wiktionary. Erudecorp ? * 20:41, 25 October 2007 (UTC)

Last time I looked, Wiktionary was a completely uncited piece of ----. I think "It's already on Wiktionary." is a non-argument. Shinobu (talk) 15:09, 23 November 2007 (UTC)

Wikipedia is by no means the only encyclopaedia to do it this way. Your suggestion would accomplish little except to make sure the pronunciations were seen by far fewer people. We should definitely keep it the way it is. Timeineurope 12:29, 25 October 2007 (UTC)

If they don't want to see it, they don't have to. Erudecorp ? * 20:41, 25 October 2007 (UTC)

Who compiled this list?

Who compiled this list, and based on what sources? You see, I came here to look up a vowel, and the first thing I see is that German "Mann" is located under "a". Well, I've been to Germany on several occasions and I know that most Germans pronounce this as ɑ. The pronunciation (the .ogg file) seems to coincide with a rather than ɑ, so at least that appears correct, but all in all such an apparent error in the first item of the list is not particularly confidence inspiring. Shinobu (talk) 15:05, 23 November 2007 (UTC)

German a varies between [a] and [ɑ]. Some editors insist that their source is authoritative and therefore that no further comment is needed. I'll remove it. kwami (talk) 15:29, 23 November 2007 (UTC)

Timeineurope, Gerbrant has a point. A lot of authorities transcribe this as a back vowel, allegedly because a lot of Germans pronounce it that way. The point of using a language in our examples in that people may be familiar with it, and if they're familiar with German having back a's, illustrating front a's with German examples will be misleading. I know you value your sources, but they aren't the only standard. kwami (talk) 19:42, 25 November 2007 (UTC)

I have yet to see a source I would consider an authority on contemporary standard German pronunciation say that this sound is closer to IPA [ɑ] than to IPA [a]. Indeed, it is often suggested that native German speakers use this sound to approximate the vowel of English cat. Timeineurope (talk) 13:07, 26 November 2007 (UTC)

Yes, when short. Langenscheidt transcribes short /a/ but long /ɑː/, while German Wikipedia says it's neither, but rather a central vowel, which would be /ä/, or if you prefer, /ɑ̈/. In fact, they use French as an example of /a/. Yes, I know we shouldn't take them to be authoritative, but they do demonstrate that variation is common, and therefore likely to cause confusion among our readers. A warning at least is in order. kwami (talk) 19:14, 26 November 2007 (UTC)

Please also have a look a bit higher up here in the section "Back and front A". I keep being confused by the allocations of these two in Dutch, German and French, which, to me, seems contradictory. In Dutch especially the distinction between /a/ and /ɑ/ is definitely phonemic with a lot of minimal pairs. The fact that usually /aː/ and /ɑ/ occur does not change this. In German the same occurs, for example <bahn>=/baːn/ and <bann>=/bɑn/. In my ears the front/back difference in much more relevant than the long/short difference. Perhaps the fact that English does not have any real /a/ and and certainly no minimal pairs contrasting with /ɑ/ makes it difficult for English speakers to make dependable judgements on these issues. −Woodstone (talk) 05:41, 27 November 2007 (UTC)

Interesting, you have the back vowel short and the long vowel central, just the opposite of Langenscheidt. kwami (talk) 06:25, 27 November 2007 (UTC)

I'd like to add that I – as a native speaker of German and a quite a trained phonetician, if I may say so – cannot make out any qualitative difference between the long /a:/ and the short /a/ in German. Neither in my own speech nor in other people's German, supposed they speak something close to High German. A qualitative difference here is merely a regional phenomenon. So for me it's phonetically [ba:n] ("Bahn") and [ban] ("Bann"). The Langenscheidt dictionaries seem to be as imprecise and misleading as they usually are. — N-true (talk) 16:18, 27 November 2007 (UTC)

In the north they say [a], in the south (Bavaria, Switzerland) they say [ɑ] – my personal impression, don't you agree, N-true? Dan Pelleg (talk) 09:30, 6 December 2007 (UTC)

Yes, indeed, southern dialects often use [ɑ], but it's not standard High German. — N-true (talk) 15:04, 6 December 2007 (UTC)

Some problems

I've got a couple of problems with some items in this list. Forgive me as a few of these are repetitions of earlier points:

1. [yː] [uː] [aː] [ɑː] [eː] [iː] [oː] [ɔː] [øː]; considering that the bottom of the page indicates what ː means, inluding these seems redundant. If it's not redundant, then this page is missing the instances of the other long vowels.
2. Similarly [ɑ̃] [ɛ̃] [ɔ̃] [œ̃] are redundant since the bottom of the page indicates what ̃ means.
3. Technically speaking, [ts] [dz] are not affricates in English. They are instead sequences of a consonant and a homorganic fricative. There are some phonetic differences. I understand comparisons to English would be helpful though. Am I being too pedantic here or should we put Like English [] in the description box?
4. [ɘ] lists Russian as being a language that has this consonant but this is, according to a source several decades old, only for some speakers and not part of Standard Russian. I notice there are a few missing central vowels from the list. Should we just remove this one?
5. [ɰ] Spanish is listed as having this sound. As velar approximant indicates, this is not accurate due to lip rounding features. Considering that [β] is listed as a Spanish example, we can move the agua example to [ɣ] and put a Japanese example for [ɰ]
6. Also, what's with the [ spaces ]? Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 07:06, 29 November 2007 (UTC)

1. I'd vote to remove the long vowels, for article length if nothing else, but several editors want them there because of the languages we're using for illustration. I think that's reasonable (there are a lot of German words in Wikipedia, and we don't need long schwa), but another possibility would be to mention differences in length in the comments section.
2. These are the nasal vowels of French. Again, they are a special case, considering the huge influence of French on English. Either that, or we could add them as marginal phonemes to help:pronunciation.
3. Since there are practically no instances of overt (tie-bar) affricates in Wikipedia, making a distinction between [ts] and an affricate may be too subtle. But a comment at least is in order.
4. Let's lose all mid-central vowels but schwa.
5. Spanish is not a good example of [ɣ], and Japanese does not have [ɰ]. We should mention that Spanish is not exactly [β].
6. The spaces are due to people trying to get things to be legible in their browser fonts. Personally, I have trouble displaying the retroflexes; the tail runs into the following bracket.

Also, you now have [ʕ] as the sound of Arabic ‘ayin, which it is not. kwami (talk) 16:36, 29 November 2007 (UTC)

a and a:

Symbol	Examples	Description
A
Open front unrounded vowelⓘ	Spanish casa, French patte, German Mann	For many English speakers, the first part of the ow sound in cow. Found in some dialects of English in cat or father.
Open front unrounded vowelⓘ	German Aachen, French gare	Long [a].

Both these are using Image:Open front unrounded vowel.ogg which can't be right. Fred Bauder 19:19, 1 December 2007 (UTC)

Why can't it be right? kwami 19:33, 1 December 2007 (UTC)

Rain??

This article states: [ ʷ ] English "rain" [ɹʷɛn] - Indicates a sound has lip rounding, "quick."

The example given, "rain," I believe, is poor. I am unfamiliar with anyone who would use "lip rounding" with this word - not that it may exist somewhere. (The example of "quick," on the other hand, is OK in my book.) Please come up with a better primary example. Charvex 07:06, 2 December 2007 (UTC)

As far as I understand, English /r/ is always labialized, especially in the syllable onset. Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 07:25, 2 December 2007 (UTC)

Personally, I pronounce quick as a cluster, [kwɪk], not labialized [kʷɪk]. kwami 12:20, 2 December 2007 (UTC)

Might be a dialect issue, though. Charvex, where are you from? GA has labialized /r/. kwami 12:21, 2 December 2007 (UTC)

I won't argue with GA. But for me, the /r/ in rain is a retroflex approximate that I would indicate with [ɻ], but this happens whenever /r/ is followed by most close mid-front unrounded vowels or the dipthong /eɪ/ for me. Pulling my tongue back to make the /r/ sound, which is almost uvular, in combination with these kinds of vowels eliminates the need for the labialization in words such as "rain." However, the labialized /r/ does occur when the vowel sound moves further back, as in a close mid-back unrounded vowel, and of course, open back rounded vowels, e.g. "rot." - - My comments are personal observations. My interest in linguistics is primarily in syntax rather than phonetics. I will defer to your better judgement. Charvex 07:21, 3 December 2007 (UTC)

I thought maybe we should specify 'American English', so others aren't confused. kwami 07:56, 3 December 2007 (UTC)

According to Luciano Canepari's Handbook of Phonetics, rain is pronounced [ɹ̱˞ʷɛ̝ˑɪ̯n̺] in British English and [ɰ̟͡ɹ̱˞ʷɛ̝ˑɪ̯n̺] in American English. His phonetic transcriptions are extremely exact, although he uses his own ideosyncratic way of transcribing. I "calculated" it back into official IPA. So yes: There's a labialization in both neutral BE as well as neutral AmE. — N-true 22:01, 3 December 2007 (UTC)

That's a good idea. The American /r/ is fairly unique. It is one of the fastest ways to identify native American speakers, and it is one of the most difficult things in that dialect for non-natives to master who wish to learn it. Charvex 22:21, 3 December 2007 (UTC)

To user N-true: Although textbooks may state neuter BE speakers say [ɹ̱˞ʷ], I think it is only spoken in the southwest part of the country anymore. Most ordinary Londoners, and nearly everyone I know in the Midlands (Birmingham, Leicester, etc.), speak non-rhotic English. Charvex 22:30, 3 December 2007 (UTC)

There isn't necessarily any connection between labialization and rhotic dialects. kwami (talk) 08:27, 20 January 2008 (UTC)

Referring here through IPA template

The German Wikipedia uses its IPA template to make all its appearances link to a page similar to this (Help:IPA) page (de:Liste der IPA-Zeichen) – immensely helpful, especially of course for non linguists, for whom it makes IPA coding hugely more accessible. How about doing that here, too? Dan Pelleg (talk) 09:43, 6 December 2007 (UTC)

We have two: {{IPA2}} and {{pronounced}}. Two others link to the help:Pronunciation page. kwami (talk) 11:49, 6 December 2007 (UTC)

We need info links

Hi, as mentioned at Template talk:Audio-pipe, we cannot remove the info links in this article if we use GFDL audio samples. Please don't delete them. --Kjoonlee 08:01, 20 January 2008 (UTC)

Mandarin tones

I'm assuming that these edits were meant for the talk page — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 07:03, 21 January 2008 (UTC) :

The listed Mandarin tones are incorrect. The following is the correct way it is pronounced and spoken in HànYǔPīnYīn:
ā=妈妈 [māmâ] "mother" High tone.
á=麻 [má] "hemp" Mid tone.
ǎ=马的 [mǎ] "horse" Fall-Rise tone.
à=骂 [mà] "scold" Falling tone.
â=妈妈 [māmâ] "mother" Light tone. But seldom indicated. Hence becoming written as: [māma]

(added by User:165.21.155.94, 22:57, 2008 January 20)

This page isn't about Pinyin, it's about the IPA, as cautioned in the examples cited. kwami (talk) 07:54, 21 January 2008 (UTC)

Subtle?

The voiced pharyngeal fricative [ʕ] is slightly comically described as a subtle sound deep in the throat. Surely a better description can be found. --NigelG (or Ndsg) | Talk 14:50, 23 January 2008 (UTC)

Yeah, it's not a good description. Care to come up with something better? kwami (talk) 16:32, 23 January 2008 (UTC)

Uh-ʔ-oh! Tricky. My first inclination is simply to delete the word subtle. The article on Ayin suggests sing the lowest possible note, then one lower: would that be better? --NigelG (or Ndsg) | Talk 18:18, 23 January 2008 (UTC)

Maybe that would help some. Either way, they're gonna hafta follow the link to have any idea of what we mean. I've also read descriptions that say if you feel like you're going to gag, you've probably got it right, because English speakers only use these muscles when they vomit. It no longer feels like gagging to me, but I think it did at first. Or maybe I just wasn't saying it right at first.

Thanks for bringing the ayin article to my attention, because it's completely wrong. kwami (talk) 18:44, 23 January 2008 (UTC)

Glad I was of some use! I'm pleased you got rid of the "bleating goat", which struck me as little better than a gratuitous piece of linguistic racism. BTW I see no reason to retain the parentheses round the final sentence: it's either relevant in its own right or should be omitted.

The Talk page claims that this article comes under WikiProject Judaism, an unduly restrictive claim IMO. --NigelG (or Ndsg) | Talk 21:10, 23 January 2008 (UTC)

PS I'm not sure the gagging is much better than the goat! In my experience the unvoiced pharyngeal fricative is easier for foreign students ("try saying hat in a loud stage whisper"). If you can manage that, you just have to voice the sound to get `ayn. But YMMV ... --NigelG (or Ndsg) | Talk 21:18, 23 January 2008 (UTC)

Voiced bilab fric

The description (Like [v], but with the lips nearly touching) should add that, unlike [v], it is not labiodental. In other words, it's less of an acrobatic feat than it sounds. --NigelG (or Ndsg) | Talk 21:38, 23 January 2008 (UTC)