Help talk:IPA/English/Archive 7

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

Archive 1

←

Archive 5

badges

We have used "badge" as and example for "dʒ" but is it not pronounced "dʃ"? I think that is how I pronounce it. —Preceding unsigned comment added by Gregcaletta (talk • contribs) 09:13, 3 August 2009 (UTC)

There are two main reasons against using dʃ, even in final position. One is that it isn't how linguists or dictionaries transcribe it and the other is that it isn't how anyone pronounces it. It's a lot more likely that you hear dʃ in your own speech because the stop element (the [t]) is very similar to /d/ in other words (like dog). Remember that English /d/ is very often devoiced. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 09:44, 3 August 2009 (UTC)

Well I would say that in "badge" the "d" is voiced but the "sh" at the end is not, but we can leave it the way it is if we're just basing it on what other linguists do. —Preceding unsigned comment added by Gregcaletta (talk • contribs) 10:01, 3 August 2009 (UTC)

It may well be devoiced to some extent phonetically, but phonemically it's certainly /dʒ/ (it's fully voiced in the plural badges, for example). And even if we were giving a narrow phonetic transcription it would probably be preferably to use the underring diacritic to indicate devoicing rather than "dʃ". +An gr 10:44, 3 August 2009 (UTC)

Hertfordshire

There's an edit war going on at Hertfordshire between an editor and an IP. After watching this go round a couple of times I decided to intervene. I looked up the pronunciation in the online OED and found it agrees with the IP. The editor accuses me of "screwing w the IPA". Since he/she also pointed at this project I hope we can get some impartial judgement on this issue. But then, I thought I was impartial 'til my edit was so rudely reverted. Bazj (talk) 07:59, 5 August 2009 (UTC)

The OED and Wikipedia have different transcription conventions, which is probably the source of confusion here. What we transcribe as ɑr the OED transcribes as ɑː and what we transcribe as ər the OED transcribes as ə. The reverse isn't always true, however. These differences allow us to indicate the distinction between the vowel of barn and the vowel of balm as well as between capper and Kappa-- something the OED doesn't indicate but that rhotic dialects make.

Looking at the edit history it seems that both the anon editor and Bazj were unfamiliar with this. Perhaps when dealing with potentially contentious editing or edit wars, we should initially point out this project in the edit summary. Something like:

Fixed IPA according to Wikipedia conventions. For details/guidelines, see WP:IPA for English

I think everyone here is acting in good faith, but with an edit summary like this, we can avoid accidentally biting people who simply don't know better. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 08:31, 5 August 2009 (UTC)

That's a good way to word it. (I do tend to get rude when the same battle is repeated over and over, when it's no fault of the new editors who haven't been through it before. I apologized to Bazj on my talk page.) kwami (talk) 09:21, 5 August 2009 (UTC)

From our discussion over at Kwamikagami's talk page I can see where you're coming from. However I have a number of reservations...

a minor point Unless you mark all your edits with a pointer to the project you can only expect to have the edits judged by standard reference works such as OED, Webster's, Chambers, or AHD. In the absence of such a pointer edit conflicts are inevitable.
Having a Wikipedia transcription convention is WP:OR. It also means that transcription's cannot be checked against any other material, which violates WP:Verifiability. (I see the question of WP:OR has been raised before, but it collapsed into a discussion of the editor's skills at IPA rather than the key issue of WP:OR).
When it comes to place names, local trumps international. Isn't that why Beijing is now preferred over Peking, and Kolkata over Calcutta?

Bazj (talk) 11:40, 5 August 2009 (UTC)

Having a Wikipedia transcription convention is by no stretch of the imagination OR. We're not deciding ourselves what the pronunciations of these places are, we're simply presenting verifiable pronunciation information in a uniform fashion. Beijing and Kolkata have nothing to do with anything; they're Latin-alphabet transliterations of the native names which have (comparatively recently) become established in English usage. Other places are still best known by their English names (Rome, Cologne, Munich, etc.), not their native names. By using a pan-dialectal transcription, we avoid ambiguity and make sure everyone's pronunciations are included; moreover, there may be more than one local pronunciation of a place. (For example, I know two people from Exeter: one is working class and calls it [ˈɛksətɚ]; the other is university-educated and calls it [ˈɛksətə]. Both are local pronunciations.) In addition, locals are not the only people entitled to utter the name of a place; Americans (for example) may want to speak of Hertfordshire too, and it would be an error for them to pronounce it [ˈhɑːfədʃə]. It would be absurd for us to give different pronunciations for, say, Jersey and New Jersey, since the two names are pronounced identically in any given dialect. +An gr 14:42, 5 August 2009 (UTC)

It doesn't stretch my imagination in any way. In the case of Hertfordshire I can go to the OED and see a transcription. That's verifiable. The Wikipedia transcription exists only on Wikipedia, hence NOT verifiable, so it's OR. Seems simple enough to me. Bazj (talk) 15:07, 5 August 2009 (UTC)

Actually, it doesn't. Go to [1] and click on "Show IPA". Or look in the Longman Pronunciation Dictionary. The OED is not actually the only dictionary on the planet. +An gr 15:16, 5 August 2009 (UTC)

Angr, At no point have I claimed OED is the only dictionary on the planet. I listed 4 in my earlier posting. The OED just happens to be my preference, and the one made available online via my local library. I'm not sure what point you're making with the link you posted, their transcription matches neither the OED nor the article.

Your link to reference.com (which I understand is the AHD) has /ˈhɑrfərdˌʃɪər, -ʃər, ˈhɑrt-/.
Merriam-Websters has \ˈhär-fərd-ˌshir, ˈhärt-, -shər, US also ˈhərt-\
The OED has (ˈhɑːfədʃə(r), ˈhɑːt-)
Longman's online version unfortunately doesn't carry the IPA.
Chambers doesn't seem to have proper nouns.

None of these matches the article's /ˈhɑrfədʃər/. The Wikipedia transcription is simply not verifiable, it's OR. Bazj (talk) 16:34, 5 August 2009 (UTC)

/ˈhɑrfədʃər/ is wrong, it should be /ˈhɑrfərdʃər/, which matches reference.com's second option and (with a trivial conversion of transcription systems) MW's second option for the final syllable. +An gr 16:47, 5 August 2009 (UTC)

Now they're all wrong? Lack of verifiability doesn't even start to do justice to this. Bazj (talk) 17:05, 5 August 2009 (UTC)

No, they're not all wrong. Kwami made a typo; he should have written /ˈhɑrfərdʃər/ instead of /ˈhɑrfədʃər/. +An gr 17:13, 5 August 2009 (UTC)

That doesn't agree with any of the 4 transcriptions in his 10 edits over the last 2 years, ˈhɑrfədʃər 3 times, /ˈhɑrtfərdʃər/ 3 times, /ˈhɑːtfʊdʃɪɚ/ once, and ˈhɑːtfʊdʃə 3 times. The latest editor's gone along with your pointer to reference.com and used /ˈhɑr(t)fərdˌʃ(ɪ)ər/. At least that one's verifiable. But I'm sure if we wait a couple of hours another editor will come along with another version.

The fact that someone who knows the subject can come up with 4 conflicting versions shows the importance of verifiability. Without it this edit war is just going to run and run. Bazj (talk) 17:53, 5 August 2009 (UTC)

What are you advocating, exactly? — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 18:17, 5 August 2009 (UTC)

I favour following the policies on verifiability, OR and consistency. In this case that means using a reliable source ~ a dictionary. If there's a transcription on wiktionary, that should be the first choice; failing that one dictionary should be chosen and used consistently. Despite Angr's view of me, I wouldn't choose the OED, because it's not freely accessible. Finally the source needs to be cited. Just what you'd expect of anything in Wikipedia really. Bazj (talk) 19:22, 5 August 2009 (UTC)

Any argument on that article should be over the pronunciation (the OED doesn't have /ʃɪər/, for example), not over the transcription. If you want to change the transcription, the place to do it is here, so that it applies to all articles equally. kwami (talk) 18:45, 5 August 2009 (UTC)

I have no problem with the pronunciation given in the audio clip. The problem lies in converting it to a verifiable, non-OR, consistent, citable transcription. Bazj (talk) 19:22, 5 August 2009 (UTC)

Sorry but in pronunciation you cannot both advocate verifiability and consistency at the same time. Phonetic transcription is inherently inconsistent in the sense it varies from publishing house to publishing house. And in this case, a consistent approach to IPA on Wikipedia wins hands down over "verifiability" (unless you're advocating some massive page listing all the divergent IPA systems used to transcribe words across Wikipedia) which I think is being abused as a policy here as it's not so much the actual pronunciation that's in question but the question of how we write it. I think we shouldn't lose track of the spirit of the verifiability policy over common sense. Akerbeltz (talk) 19:42, 5 August 2009 (UTC)

"it's not so much the actual pronunciation that's in question but the question of how we write it" ~ we are in agreement on that point.

"it varies from publishing house to publishing house" ~ so if the same publisher is consistently used as a source we get consistency and verifiability.

Verifiability should not have been ditched in the quest for consistency. Nor should the prohibition on OR. A pre-existing scheme should have been settled on and used consistently. That would satisfy the rules on verifiability, OR and consistency. Bazj (talk) 21:17, 5 August 2009 (UTC)

For that approach to work, you'd need a single publisher that has IPA for every word that's in question here. Since such a source does not exist, clearly that's not a workable approach then? Akerbeltz (talk) 21:21, 5 August 2009 (UTC)

Using such a source (despite its limitations) allows the IPA-illiterate to contribute in seconds. The WP:OR "Wikipedia Transcription Method" reduces the available pool of editors to those:

who have some understanding of IPA,
who have knowledge of the pronunciation of the word in question,
who are willing to set aside the IPA methodology they know,
who have the time to read up the Wiki method, and
the have the time to transcribe the word.

This is a vanishingly small pool of editors.

"Since such a source does not exist" - Reference.com from what I've seen has pretty solid coverage. Bazj (talk) 14:19, 6 August 2009 (UTC)

And I still think quoting the verifiability guideline for IPA transcription of words with a known pronunciation is spurious. We're not debating the veracity of a fact, just how to spell it out. I mean, if there was an argument over it being a velar vs a dental fricative, then yes, verifiability by all means. But here?? And incidentally, I don't think any of the printed ancyclopedias "verifiy" their IPA en masse, maybe in some controversial cases but overall, they'll just apply their in-house style across the book. Akerbeltz (talk) 21:43, 5 August 2009 (UTC)

Wikipedia's rules don't ask us to verify other encyclopedias, merely to use them as verification for what's written in wikipedia. Bazj (talk) 14:19, 6 August 2009 (UTC)

It's a naive assumption that an IPA illiterate or near illiterate editor will contribute in a helpful way using such a source. I have taught phonology related topics to the "masses" for decades and you'd have a hard job a) convincing me that IPA illiterate people even notice the IPA and b) are willing to touch it in any shape or form. And such contributions would have to be checked by a third party in any case because you and I may know the difference between e ə and ɘ when we see it, but most people will end up entering the wrong one and then people like kwami can trawl through hundreds of pages checking it was done right... This is one of those topics where less haste is more speed by relying on people literate in IPA and phonetics enough to be able to adjust whatever source to the "Wiki standard". Akerbeltz (talk) 14:33, 6 August 2009 (UTC)

This being a wiki, there's no requirement that you add pronunciation information in picture-perfect IPA that conforms to Wikipedia's transcription convention, merely that you not edit-war about it when someone else tidies up your pronunciation information so it does conform. +An gr 14:36, 6 August 2009 (UTC)

There lies the problem with the WP:OR Wikipedia transcription convention. When you have an IPA-literate editor putting on the WP IPA with no verifiable source, and an IPA-illiterate editor who has picked up a transcription from a dictionary/encyclopedia with an impeccably verifiable source, then conflict is almost inevitable. The more I think on it the more I'm surprised there aren't more edit wars on IPA transcriptions. Bazj (talk) 14:51, 6 August 2009 (UTC)

That's an impersonal assumption of bad faith. The main reason why an edit war would start is because editors might not realize our transcription system is different and why. This is why I recommended the edit summary above. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:46, 6 August 2009 (UTC)

That's because most people don't have a problem with the system advocated/used on Wikipedia... Akerbeltz (talk) 16:40, 6 August 2009 (UTC)

More likely they view it as irrelevant to their needs and incomprehensible. Bazj (talk) 16:49, 6 August 2009 (UTC)

Don't we have a help page somewhere that aids in seeing the differences in pronunciation indicators accross dictionaries and whatnot? — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:46, 6 August 2009 (UTC)

Do you mean Pronunciation respelling for English?

Yeah, that's what I was thinking of, though it's not in help: or WP: space. Perhaps we can create a help/WP page for editors like Bazj (and to a certain extent, myself) who may need help translating the information they get from dictionaries into Wikipedia's IPA transcription scheme. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 23:20, 6 August 2009 (UTC)

That has two dict. IPA schemes on the left. We could copy those and add more. One problem, however, is when the dict. uses multiple transcriptions. It may be difficult to tell when that means there is actually more than one distinction pronunciation, as opposed to regional variation. So, for example, dict.com frequently lists both /ɔr/ and /ɔər/, which only reflects the fact that many speakers do not distinguish horse from hoarse, not that speakers who do make the distinction vary as to which they use. kwami (talk) 00:53, 7 August 2009 (UTC)

Yeah, we should make it clear when a dictionary doesn't make a certain distinction. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 05:04, 7 August 2009 (UTC)

I started a draught guide here, so far not linked to anything but this talk page. kwami (talk) 08:29, 7 August 2009 (UTC)

So what you're really saying is that you're the only one who has an issue with it? Akerbeltz (talk) 21:45, 6 August 2009 (UTC)

If you're blind to all the other times it's been brought up, then yes, it's just me.

Bazj (talk) 11:04, 13 August 2009 (UTC)

The original objection was having /r/ in the name of a place where people have a non-rhotic accent. This hasn't been much of a problem—occasionally s.o. gets upset about 'cultural imperialism' or the like, but for the most part Brits seem willing to accept it. (And of course we can always add a non-rhotic local pronunciation if need be.) This is a more tolerant attitude than I expected at first, and in marked contrast to /ju:/ after /t, d, n/ in US place names, which tends to make people go berserk. kwami (talk) 22:26, 6 August 2009 (UTC)

Bazj, have you read the listed conversations? All but the second (and to a certain extent the fifth) are objections to using IPA at all, not to our synthetic polydialectal approach. One other person brought up your same concern, though eventually focused more on its phonetic unreality. If I understand correctly, yours is a concern of accessibility, which I believe we're attempting to address. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 14:12, 13 August 2009 (UTC)

Optional phonemes, continued

Picking up on the discussion above, if indeed this page is still open to proposals, I'd like to chime in. I had a recent experience where a columnist wrote a piece excoriating my transcription of his name. If I could respond (he's not Wikipedia-savvy), I could direct him to this page or explain stuff about accents and transcription and so on. I might even suggest a [phonetic] transcription beside the /quasi-phonemic/ one. (I verified on YouTube that he is non-rhotic (and TRAP-BATH split, etc.).)

This does support the superscripted-r idea. Whatever other meaning it has in the IPA is not really relevant; an in-house system can shake things up if need be (there's consensus on this point now I think). The best argument against it is that (according to Kwami) it mostly hasn't been a problem; a comment here was that even London placenames are transcribed rhotically without complaint.

Anyway, here is a proposal for the yod: transcribe as /ɪu/ instead of /ju/. My arguments:

Many accents use this very pronunciation

It does feel more like a diphthong than a consonant cluster. E.g., /stj/ is an odd cluster, occuring in no other context.

/j/ means something on its own, whereas /ɪu/ is funny linguistics stuff people will have to look up before objecting, and this page will tell them what they need to know. /nɪu jɔrk/ may be more palatable (although I don't know why it needs a pronunciation guide...), and /kɪut/ isn't hard to understand.

Right now it seems RP decides if the yod is in, but once we're poised against complaints, we can extend to other contexts, e.g., blue, chute, chew, rude, abstruse. Why not? (Some yods will get missed, but in cases where few pronounce them anyway.) Word-initially, it's probably best to stick to /ju/ (maybe /jɪu/).

What do you all think? I reali(s/z)e it's not a perfect proposal.

One more comment: /ɔər/ just bugs me. The other "breaking" diphthongs can be justified by actual (widespread) use, but this one? /oːr/ would confuse no one and have an immediately clear pronunciation in all dialects, which is not the case with /ɔər/. Also, this vowel is different from the others; in other cases the "short" version (/ɪr/, etc.) is almost only pre-vocalic, but not /ɔr/. Finally, a lot of the displeasure in the archives cites this one part of the system, and no doubt there is more unexpressed displeasure out there. (And, I'm not sure it's even being used; George Soros can't be right.)

Tyuia (talk) 22:13, 8 August 2009 (UTC)

I personally like the superscript r. All that would mean in IPA is something like 'r-colored', which we might be able to argue successfully. At the ends of words it's simply /r/ phonemically, but it might be a wise concession to write it superscript there too. Would we only have such transcriptions for non-rhotic personal and place names?

If we're going to go that route, I'd suggest superscript j for the analogous situation in /nju: jork/.

I don't get the point of /iu/. If we're going to transcribe ewe /ju:/, why not be consistent with cute /kju:t/?

/our./ is more variable in its dictionary conventions than most r-colored vowels. That is certainly something we can discuss. It would also be a fairly simple task to go through all IPA-en transclusions and replace it systematically with AWB if we decide on a change; we'd just want to be sure a good majority is on board before we do that. kwami (talk) 06:56, 9 August 2009 (UTC)

Somebody mentioned above that superscripted r has another meaning in IPA, but I cannot find any evidence of that in the IPA handbook. So we would be free to use it for an optional r. Similarly we could do use supercript j for optional /j/ before /u/, since it is close enough to the meaning of palatalised that it has in IPA. This would leave us with a reasonably intuitive /n^juː ˈjɔ^rk/. −Woodstone (talk) 08:04, 9 August 2009 (UTC)

No, it has no defined IPA usage. However, any IPA letter can be subscripted to signify a 'coloring' of the preceding letter. So, for example, superscript schwa or t is often used for epenthesis ([dæn^ts]), superscript j or w for diphthongs (English [iʲ] and [uʷ]), s or r for fricative or trilled release, etc. An "r-colored vowel" use would be in line with that convention. kwami (talk) 08:49, 9 August 2009 (UTC)

I also like the idea of using superscript to indicate optional /r/ and /j/ used in r-coloured vowels and yodding — eg, New York would be represented as /n^juː ˈjɔː^rk/ and New Jersey as /n^juː ˈdʒɜː^rzi/. – Marco 79 15:05, 17 August 2009 (UTC)

I find both superscript j and superscript r problematic. With superscript j, we're basically marking a distinction that speakers are unable to discern themselves and that dictionaries don't encode for. We'll thus find so many cases where the superscript j is misapplied or where a nonsuperscript j appears when it should be superscripted, that the distinction would be meaningless.

It's even worse with superscript r, as the only time it would appear would be in the syllable coda and non-superscript r would never appear in the coda. We would thus be making the system more complex just to indicate something that is already apparent to anyone who reads the key (something they would have to do if they don't know what the superscript r means). — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 03:15, 18 August 2009 (UTC)

Then, fear would need to be /fɪ^ər/, as serious–Sirious-merging speakers would omit the ə and non-rhotic speakers would omit the r. And that would start being really ugly. And while we're at it, we could write /d^æɑːns/ and /^ɒɔːf^ən/, or even /f^ʌɜr/ ~ /f^ɪɜr/ ~ /f^ɛɜrn/ and let everyone drop what they will. ;-) -- A. di M. 09:14, 18 August 2009 (UTC)

I strongly oppose the use of superscript j for optional /j/. It has a well-known different meaning in IPA, namely palatalization. There already is an established standard for indicating optional phonemes in IPA, namely to put the offending phoneme in brackets: /n(j)uː ˈjɔː(r)k/. Why don't we use it too, instead of inventing our own system which is even incompatible with IPA? — Emil J. 10:22, 18 August 2009 (UTC)

Because it's not "optional" in the sense that speakers freely alternate. In most words, in each given dialect the yod is either forbidden or mandatory. As for "incompatible with IPA", this is a (pseudo-)phonemic transcription, not a phonetic one, so the phonetic meaning of the symbols shouldn't be taken too seriously; otherwise you'd believe that the first vowel in "burner" is more open than the second one, whereas if there's a definite difference in height between the two at all, it's the other way round. --___A. di M. 10:48, 18 August 2009 (UTC)

Superscripting is an awfully big solution if it's not a problem. I'd wait till rampaging mobs of RP speakers demand it; otherwise we have to change to /dɛnvə^r/, /pɔə^rtlənd/, and thousands more, which seems silly. The yod problem, on the other hand, is really causing edit wars.

What is an "optional /j/"? /bju:ti/? /strju:n/? Or are these required and disallowed, respectively?

The point of /ɪu/ is, among the reasons noted, that /nju:/ sure looks like it's sounded /n/ + /j/ + /u:/, and that's what throws people. But a diphthong cannot be expected to yield to such decomposition; I don't think /oʊ/ matches any major dialect. A variety of reasons may explain why the /r/ doesn't annoy non-rhotic speakers as much, such as it never being pronounced in such a position (cf. /nu: mɛnju:/), and it being spelt thus in plain text, and it being sounded in dialects familiar to them.

The reason for the inconsistency is that you and ewe are homophones almost everywhere, including I believe East Anglia, which drops yods like crazy. Call it a compromise of purity with the facts on the ground.

Does anyone find /fɔərd/ easier to understand and more sensible than /fo:rd/ (or /ford/)?

Tyuia (talk) 09:25, 9 August 2009 (UTC)

'Optional' j is j in ju: after d,t,n, etc.

/ɔə(r)/ is the OED convention, so it's pretty well established. Best to avoid the five basic vowels, as in /or/, because they're so ambiguous. kwami (talk) 10:21, 9 August 2009 (UTC)

"Menu" is two syllable /mɛn.juː/, so you'd pronounce the /j/ much like you'd do in "unused" /ʌn.juːst/. --A. di M. 00:12, 11 August 2009 (UTC)

I'd assume /ə'sju:m/, /ɪn 'lju: əv/, but not /flju:t/ or /sju:t/ or /ɪsju:/? OED uses /ɔ:/, does it not? /ɔə/ was from when London was less NORTH-FORCE merged and the vowel was pronounced that way. K&K use /or/, so it also has precedent. Is /o:r/ also ambiguous? I'm not sure I understand your concern there. Tyuia (talk) 16:50, 9 August 2009 (UTC)

/flu:t/, but /s(j)u:t/ and /ɪʃu: ~ ɪsju:/. I assume that means that /nju:/-pronouncers vary in their pronunciation of suit.

OED2 used /ɔə/. OED3 has dropped the distinction, and so is irrelevant—cept for their US transcriptions, which use /ɔ(ə)r/. The problem with /or/ is that in general we will have no idea if an editor meant /ɔər/ or just /ɔr/, given the ambiguity of /o/. kwami (talk) 23:21, 9 August 2009 (UTC)

/sju:t/ and even /lju:d/ are fairly regressive, /kən'sju:m/ and /lju:/ not so much. Anyway, a superscript-j has the problem that it looks like part of a palatalized consonant (whereas /ɪu/ looks like a vowel).

I still don't see the problem with /or/. With /ɔr/, we have no idea if the editor meant /ɔər/ (see Soros above). /or/ is not ambiguous if we defined it unambiguously. In any case, not understanding your objection, I'm not sure if it applies to /o:r/, which I still find much better than /ɔər/.

However, I don't have the patience for one of those protracted Wikipedia campaigns, so if no one else agrees then I give up. Tyuia (talk) 04:25, 10 August 2009 (UTC)

How 'bout this? --A. di M. 00:08, 11 August 2009 (UTC)

Maybe that will help, but there's already a footnote explaining this. The big problem may be people rebelling against transcriptions before they RTFM. Tyuia (talk) 04:52, 11 August 2009 (UTC)

As for the exception of 22:13, 8 August 2009, above: /ɔr/ is not "short" the way /ɪr/ is. It is the rhotic counterpart of /ɔː/ (like /ɑr/ is that of /ɑː/); the "short" one is /ɒr/. I would have no strong objection to transcribing them as /ɑːr/ or /ɔːr/. The only drawback is that we'd need syllabe breaks to show that "orring" (the gerund of the verb meaning "to perform a bitwise OR operation with") doesn't rhyme with "awe ring", or that "bra ring" doesn't rhyme with "starring"; but that'd be a very rare issue, as the things we transcribe as /ɔr/ and /ɑː/ are very rare at the end of syllables. But I think that the current system ain't broken and doesn't need to be fixed.

As for the proposal of using /ɪu/ wherever it might be distinguished from /u:/ including chew or rude, it'd be useful for Welsh English, where (if I understand correctly) there's a distinction between you /ju:/, ewe /ɪu/, yew /jɪu/; but we aren't trying to trascribe all dialects in the world no matter how minor (and I wouldn't be able to imagine a clear way of transcribing the fur/fir distinction without interferring with the possibility of showing that "stirrer" doesn't usually rhyme with "mirror"); so I'd say that where neither RP nor GenAm nor StdAus use the yod, we can omit it as well. And when it won't ever appear right after /tʃ/ or the like, there's no serious problem in transcribing it with /j/. IOW, I think that the current system ain't broken and doesn't need to be fixed. --A. di M. 13:34, 11 August 2009 (UTC)

Well the battling on New York suggests the current system can be improved, as does the mistranscription of FORCE vowels. That's Wikipedia-- constant refinement. And, /ɔər/ is just ugly and reflective of almost no one's pronunciation.

/ɔr/ has two main sources: from non-pre-vocalic /ɒr/ (north) by lengthening, and from the historic sound in aural, centaur, etc., which I think are all classical imports and less common. These two cases are in almost complementary distribution (centaur being about the only non-pre-vocalic case) and contrast in only marginal dialects (any?), so there is little point in distinguishing them. Your awe ring example is expectedly contrived, and the word break covers that case. The only troubling case I can think of is abhorrent. Conclusion: it's halfway a length distinction :).

Anyway, what you say partly supports my point. The contrast between NORTH and FORCE is not analogous to the other schwa/schwa-less (tense/lax) examples, and so we'd be justified in not following the same pattern. Hence, I propose /ɔr/ for NORTH and /or/ or /oːr/ for FORCE. If you want /ɔːr/ for NORTH, that's fine, but I'd follow the example of /ɑr/.

Yod-dropping is a very variable process. Yes, few dialects say /skrjuː/, but among real-life RP and GenAm speakers there is so much variation that I don't see the harm in notating a large variety of historic cases rather than drawing a bright line. It helps that it's pretty easy to do so based on spelling, and (unlike the example of fir/fur) it doesn't cause any systemic problems. Tyuia (talk) 20:16, 11 August 2009 (UTC)

Actually, "awe ring" was a (minor) point against /ɑːr/ and /ɔːr/, which I mentioned as the only drawback I can think of such a transcription. (It'd not be very serious, because we can use a period for syllable breaks: /ˈɔːr.ɪŋ/, /ˈɔː.rɪŋ/, /əbˈhɔːr.ənt/. For an articulatory POV, the main difference is that /r/ is rounded in onsets and unrounded in codas: [ɔːɹɪŋ], [ɔːɹʷɪŋ].) I have no opinion about /ɑːr ɔːr/ vs /ɑr ɔr/, though the first one is that I would have used if I had invented this system myself. As for yods, I'd draw the line excluding the ones after tʃ, dʒ, j, ɹ, Cl (where C is any consonant), and including the others. YMMV. (The point is, very few dictionaries would transcribe the difference in such places, where not even the most old-fashioned flavour of RP would use it, so we'd have to guess according to the spelling and that'd be WP:SYNTH.) -- A. di M. 10:26, 12 August 2009 (UTC)

WP:SYNTH is a big hammer for whacking a pronunciation guide; most are now not sourced at all, and to any with knowledge of English's phonological history it is usually a "routine calculation", especially if etymology is available. There are also old dictionaries, cognates, and native speakers who make the distinction. I still like /ɪu/, but what small number of people are on this talk page have shown no enthusiasm (/ɨnθɪuziæzəm/) for it.

Another problem with /ɔər/ occurred to me. For many speakers, such as nearly all Canadians and many Americans, /ɒ/ and /ɔ/ (in this system) are realized as /ɑː/, while /ɒr/ and /ɔr/ are more like /or/. That's a pretty big mismatch between symbols, but /ɔər/ is even worse; imagine trying to sound out AH-UH-ER for the vowel of FORCE! /or/ or /oːr/ per my proposal is (close to) exactly right for these speakers, and will cause no difficulty for those who use /ɔr/ in these positions, as "or" is how that sound is written in plain text, and they do not have /o/ + /r/ in any position (low-rider or whatever don't count for the usual reasons).

In fact, this whole system looks like just RP + rhotacism. And we get complaints about the /r/s! Hm....

Tyuia (talk) 01:46, 13 August 2009 (UTC)

/'fɔərs/ for force is just as close to the US pronunciation as /'pɪərs/ for pierce. In fact, the OED has [ɔ(ə)r] as the modern US pronunciation. If people ignore the schwa in /'pɪərs/, they can ignore it in /'fɔərs/. There's nothing wrong with /'foːrs/, but just about the same argument could be made for /feːr/ for fair. kwami (talk) 05:11, 13 August 2009 (UTC)

My entire argument above is how /ɔər/ is not analogous to /ɛər/. This last point: The chart gives /ɔː/ as the vowel in THOUGHT. For cot-caught+father-bother merged speakers (close to half of North Americans), this will be [ɑː], and FORCE is nothing like [ɑər], so that the same symbol is being used for totally different sounds. In contrast, /ɛ/ as in DRESS plus /ər/ is reasonably close to what nearly all North Americans use for SQUARE.

The other points in summary: (1) RP speakers, who are served by the vowel+schwa notation, do not distinguish NORTH and FORCE. (2) The other cases pair into shortish/longish and only-pre-vocalic/all-environments, but these don't; e.g., the word NORTH. (3) /or/ (or /oːr/) will for RP and other many speakers immediately read as /ɔː(r)/, because in written English that's the sound it represents (not true of -er-, nor really of -ir-). (4) (Newish point.) As I understand, unmerged American speakers use a different vowel quality for NORTH and FORCE, rather than a schwa.

Tyuia (talk) 07:57, 13 August 2009 (UTC)

Weighing in on /ɔər/. My understanding of the system is that it's largely a compromise between RP and GA, the accents that are represented in most dictionaries. When it comes to vowels followed by r, the system seems to very often take the RP representation of the vowel (that is with a centering diphthong) and then add the r. /ɔər/ is an exception to that. According to our article on Received Pronunciation, RP /ɔə/ and /ɔː/ merged to the latter in the middle of the last century. Thus, rather than having the RP-vowel + r, we have some-anomalous-anachronistic-diphthong + r.

So if we represent the rhyme of beer as /ɪər/ instead of /iːr/ because Brits would object for POV concerns, I suspect that there would not be the same objection of using e.g. /oʊr/ over /ɔər/. We can also easily argue against using the RP-vowel + r in this regard (/ɔːr/) because RP does not make this distinction. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 14:47, 13 August 2009 (UTC)

I really wonder if we even need to distinguish FORCE and NORTH. If our intention is to cover primarily RP and GenAm, we don't need to, since the distinction is sharply recessive in both dialects. I think Scotland is the only place where the majority of people still distinguish "horse" and "hoarse", and this page otherwise doesn't bother with distinctions that are found only or predominantly in Scotland. +An gr 15:03, 13 August 2009 (UTC)

Readily available reference works, such as most American dictionaries, make the distinction, so it's easy for us to do the same. Covering RP & GenAm should be a minimum not a maximum requirement. Scottish, Australian, and Canadian English are certainly important dialects, but many of their distinctions aren't well-documented. Plus, they might cause systemic problems (mirror/stirrer), which is also not a problem with NORTH/FORCE. Tyuia (talk) 18:42, 13 August 2009 (UTC)

Yeah, as much as we like to call this "pan-dialectal", it's probably more accurate to call it multi-dictionaryish. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:58, 13 August 2009 (UTC)

A pan-dialectal transcription would be very tricky, and would involve original research. From the top of my head I couldn't name a reliable source telling whether "colonel" has the fir vowel, the fur vowel, or the fern vowel; or whether gauge has the days vowel or the daze vowel; or whether axion has the bad vowel or the lad vowel. OTOH, each of the distinctions we currently make is made by most British dictionaries, most American dictionaries, or both. -- A. di M. 20:40, 15 August 2009 (UTC)

Insane edit-warring

Two users are edit-warring against me to remove an additional example of a distinction not covered, one that it might be helpful to users to mention. Nothing they say in their edit summaries makes any sense at all. The example was the distinction in some accents between the vowels of tight and tied. I picked this pair because "Canadian raising" separates them, as does Scottish English's similar raising phenomenon. (Just as I picked bad and had for short-a splitting because those two have distinct vowels in Philadelphia, New York, and Australian dialects, whereas (e.g.) bad-lad would be the same in New York, sad-had would miss Philadelphia, etc.)

First, I'm told this distinction is "not phonemic". This is patently false, as anyone familiar with the subject can tell you. Common minimal pairs are writer/rider (for "flapping" raisers) and even idle/idol for some. Then I'm told it's conditioned by a following voiced consonant. This is also false; e.g., the vowel in spider is often "Canadian" raised. See J. Fruehwald, "The Spread of Raising: Opacity, Lexicalization, and Diffusion" for more examples and surveys that show beyond doubt that the distinction is phonemic for some. The Scottish vowel phenomenon distinguishes tied and tide similarly and again is clearly phonemic.

An editor demands that I produce minimal pairs. I do so. Another editor then insists that I must use such a pair to explain the difference in the text! He makes that change himself. But a minimal pair, though a "proof of concept", is not always the clearest example of a difference, and since I was trying to cover multiple accents with one example, I chose one suited to that goal. The minimal pairs covered one accent type each.

Then this editor completely reverts himself and me, saying that we already cover the distinction! ???????? The table plainly uses both write and ride as examples of /aɪ/.

I feel like I'm in a lunatic asylum.

Tyuia (talk) 22:58, 13 August 2009 (UTC)

If it's phonemic, you should be able to provide examples. Also, that sentence is describing things that aren't covered by this key, while writer-rider and tight-tide *are* covered. Idle-idol may be an example—I'm not familiar with it, despite making this distinction myself—but that's not the example you used in the text. kwami (talk) 23:22, 13 August 2009 (UTC)

What did I just say? I provided examples, I provided a reference, it's not covered by the key. No, I used a different example, for reasons I just said. Tyuia (talk) 01:22, 14 August 2009 (UTC)

The writer/rider tight/tide distinction we make is between /t/ and /d/. Is there a dictionary that marks the /aɪ/ /ʌɪ/ distinction? — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 02:57, 14 August 2009 (UTC)

I'm hoping that the edit-warrers weren't so daft as to not read that I wrote "the vowels of tight and tied" (both in my version of the key [2] and in my comment at the start of this section). Maybe it's not surprising, as kwami does not appear to read anything I write, but just asks me over and over questions I just answered. Next he'll be removing fern from the intro because of the -n; in fact bad, lad, and had are distinguished by the first consonant. Wow!

To answer your question, I believe no dictionary marks this distinction. That's why I (sensibly) wrote in the key that we don't make this distinction because it's not made in dictionaries. Apparently this helpful tidbit of information is like poison to the WP:OWNers of this page. Tyuia (talk) 03:14, 14 August 2009 (UTC)

It could be argued, perhaps, that such a distinction is not even expected to be made since no dictionary marks it. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 03:25, 14 August 2009 (UTC)

The same could be said for bad-lad, maybe fir-fur-fern, etc., but there's no harm in letting readers know that vowels that seem distinct to them are not notated differently in this key, and why. The tight/tied distinction is made by a large number of people, I expect far more than the other two put together. Tyuia (talk) 03:29, 14 August 2009 (UTC)

All right, two things. First is that I took the time to read the Fruehwald article and it's not as clear-cut nor as common as you state. For example, the idle/idol distinction is reported in only one informant. The gyst of Fruehwald's thesis is that the distinction may be "lexicalized" which can make for a future phonemic split. A phonemic split (or lexicalized contrast) between /aɪ/ and/ʌɪ/ is not common (especially in comparison to an allophonic distinction) nor apparent to those who are most likely to be making such a contrast or "anyone familiar with the subject." He even points out that the rider/writer distinction is "a typical example of phonological opacity" (that is, the ordering of abstract phonological rules that precede phonetic output) and is thus not a minimal pair (in the sense that it indicates a phonemic contrast) for those who pronounce the two differently.

Secondly, your tone and your accusations of bad-faith editing are in poor form. If you don't feel you can discuss this issue in a calm, civil, and constructive way, then I suggest you take some time to cool off. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 04:12, 14 August 2009 (UTC)

The paper says explicitly, "The presence of a lexical effect upon raising, even after controlling for the phonological environment seems to indicate that Raising has phonemicized." Kwami himself says he distinguishes idol/idle, and the spider/rider example is telling. Certain words are more likely to be raised than others, based not obviously on environment, and many speakers consistently assign them to different categories. All I said in the text was that some people distinguish them, not a very radical claim. Even if it were "just" allophonic, what is so bad about noting that we don't cover the distinction, that the only just response is to revert, revert, revert on sight.

This phenomenon has unfortunately had but a small number of academic studies, but they give consistent results across many North American groups, including the unpredictability of raising, so yes I'd say it is common.

I know my tone sounded harsh, but my narrative above should makes it clear why. You can understand my experience of having my edits repeatedly stomped on with the most irrelevant justification. Someone says, "If it's phonemic, you should be able to provide examples", in response to a comment that lists three examples, and says that the distinction I noted was not the one I used in the text, when it's transparently the same phenomenon (PRICE splitting/raising), and claims we already distinguish two vowels which we obviously don't. So, yes, I snap back when treated like this. If someone revert-wars while not even acknowledging the other persons's comments, they should not expect Ghandiesque patience. I dispute the claim that my comments have not been constructive, as my summary provides a good outline of the issues.

Feel free to also chide Kwami for edit-warring and not engaging in discussion with what I actually said. Tyuia (talk) 05:22, 14 August 2009 (UTC)

We criticized you for not giving phonemic examples in the text, and you did not correct that when you restored you edit, so our criticism stands. You have done better here on the talk page. I do not distinguish idle-idol myself, but I do distinguish spider-rider, another example which just popped up here without being included in the text. I have nothing against saying that we don't distinguish these vowels, but I am opposed to implying that we would not distinguish writer from rider or tight from tide, when we obviously would distinguish those words. kwami (talk) 05:40, 14 August 2009 (UTC)

I gave wonderful examples in the text. You have no valid criticism. If you would take time out from revert-warring to read a single word I wrote, you would notice all the times I said "the vowels of tight and tied", starting with the text you erased over and over again without apparently reading it and followed by twice on this very talk page. (In fact, I could barely believe it when the other editor suggested that you were talking about the whole word, because that would be such an epic oversight. I'm still waiting for you to complain about the n in fern.) Tyuia (talk) 06:09, 14 August 2009 (UTC)

We have been reading what you're writing, and you still haven't shown that the vowels of "tight" and "tied" are phonemically different in any dialect, since they don't occur in the same environment. My point was that the difference in the vowels is conditioned by the voicing of the following consonant, so the distinction in the vowels isn't phonemic, any more than the difference between the vowels of "pick" and "pig" is phonemic. The Scottish example of "tide" and "tied" is also not clearly phonemic, as you claim, since the words are morphologically different. "Writer" and "rider" are phonemically distinct in the consonants even if the only surface distinction is in the vowels; "rider" and "spider" are morphologically different again. "Idol"/"idle" is the only example you've given where the distinction is clearly phonemic, but if only one person has ever been recorded as having it, it's hardly a characteristic of an entire dialect. (Similarly, I know one Wikipedian who claims to have different vowels in "hide (v.) = conceal" and "hide (n.) = skin of animal", but if he's the only one, it's not characteristic of an entire dialect.) +An gr 09:04, 14 August 2009 (UTC)

The n of fern does not condition the distinction. The voicing of tight does. You of course know this, so I wonder why you're being so obtuse. If you had given an example good enough to illustrate the difference to anyone who didn't already know about it, say spider-rider, I doubt any of us would have had a problem. Instead, all we'll get is readers pointing out that the difference between tide and tight is the final consonant, not the vowel, so what in the world are we trying to say? kwami (talk) 08:53, 14 August 2009 (UTC)

If the informant is reliable (in these studies people are often not even told their accent is being examined, so no fudging), then one speaker is a lot, as language is not an island. It was a fairly small survey, too.

You asked for a minimal pair. I gave you one, now you say that's not enough, on the theory that there is underlying voicing that is deleted in realisation. I think this theory is dubious, and offer as one piece of evidence the occasional hypercorrection by flappers--e.g., saying lady with a t--which indicates there is no underlying form left. But that aside, what would you accept? If some distinguish bridle/bridal, you could say it's the -al suffix, etc. Maybe idol has an underlying /gsjrk/ sound that is deleted in realisation.

On this very page someone asked the difference between bad and lad; if one is unfamiliar with that accent, it might seem kind of odd. I don't think anyone reads that and is befuddled, noting that the difference is of course that one starts with a b. Similarly, if we note that in some accents the vowels of tight and tied are different, no one will think we're talking about the consonant, though they may not understand what the distinction is. Maybe type and tied would be better.

You still seem to be identifying illustrating the distinction with proving it. Minimal pairs are good for the latter, but any pair with the right sounds will do for the former. There is no need for them to be minimal; indeed, the -ad words are not. Tyuia (talk) 10:33, 14 August 2009 (UTC)

There are doubtless many pairs of vowels which are distinguished in some dialects but not in the standard ones - there's no need to list all of them. I think we have enough examples in that paragraph already (though perhaps the words " - for example - " should be added).--Kotniski (talk) 08:32, 14 August 2009 (UTC)

PRICE raising is a widespread phenomenon; if we can mention bad/had, why not that too? Anyway, it's not like it's absolutely vital it be immediately removed from the page. Tyuia (talk) 10:33, 14 August 2009 (UTC)

There are four editors who have criticized you in this. Don't you think that maybe, just maybe, you've failed to demonstrate your point? You haven't even attempted to take our concerns into account. kwami (talk) 10:49, 14 August 2009 (UTC)

Maybe you missed the four paragraphs I wrote above. And this "criticism" (we have enough examples) is pretty different from the ones you have made (the distinction is not phonemic). Tyuia (talk) 10:59, 14 August 2009 (UTC)

PRICE riasing is widespread, a phonemic distinction between /aɪ/ and /ʌɪ/, which is not necessarily the same thing, appears to be both rare and unapparent to the very speakers who exhibit it. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 18:30, 14 August 2009 (UTC)

The surveys quoted indicate it is not so rare. "Unapparent" is a bit relative-- many are not too reflective about their speech-- but if asked such speakers will notice that spider and rider do not rhyme perfectly. Tyuia (talk) 19:45, 14 August 2009 (UTC)

Really? On page 5 Fruehwald states (referring to another author's interpetation) "...children at a young age may have a phonemic contrast... but once they begin to perform the proper morphological analysis, the distribution of the two phones will become entirely predictable..." Fruehwald calls this interpretation "questionable." Furthermore, the number of papers suggesting (not arguing) that raising is lexicalized (not phonemicized) are few.

There's also this gem: "if [ay] ~ [ʌy] were to phonemicize, it would in all likelihood, happen silently until their distributions began to change as a result of other language changes." This is pretty strongly saying that it has no phonemicized. On page 18, he says "The presence of a lexical effect upon raising, even after controlling for the phonological environment seems to indicate that Raising has phonemicized" amongst his informants. I understand he's making an argument, but it's not a very strong argument.

Most importantly, and I don't know why you're ignoring this, lexicalization is not the same thing as phonemicization. Here's another way to look at it. If it is indeed the case that the number of words with Raising has increased to different environments ("diffused"), then that's a clear indication of not being a phoneme. It's equivalent to saying that the glottal stop is a phoneme separate from /t/ in Estuary English because it is no longer limited to word-final contexts. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 23:48, 14 August 2009 (UTC)

Re the gem: Surely your reading of it is wrong, as witnessed by the author's later conclusion that it has phonemicized. The "were to" refers to, what if this explanation is correct. The cited paper (Mielke, Armstrong, and Hume) takes it as given that the distinction is phonemic and tries to explain the distribution knowing this. Another good minimal pair they give is Eiffel/eyeful. They propose a theory that "while the vowels are no longer in allophones, the artifacts of their previously allophonic distribution still linger in the lexicon." Note the dependent clause.

Hayes's theory described on page 5 is in reference to a theory that "doesn't have need for phoneme inventories"; in normal models, Fruewald says, that doesn't really work.

This is not the same as "glottalstopisation" in certain environments. I'd argue that the vowels of TRAP and BATH are different phonemes in the relevant accents, by noting unpredictability (e.g., ant/aunt as a minimal pair). Or short-a tensing as phonemic by noting that planet in Philadephia is pronounced tensed even though -anV- tokens are usually not tensed. It would be pretty out there to suggest this is non-phonemic "diffusion".

Anyway, I'm gratified we're actually talking about the issues now, instead of the irrelevant unpleasantries thrown at me initially (not by you). Tyuia (talk) 02:17, 15 August 2009 (UTC)

You lost me. How is arguing that the spread of short-a tensing is diffusion "out there"? Wouldn't unpredictability be the marker of diffusion? — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 04:21, 15 August 2009 (UTC)

And Tyuia, you're still insisting on your really bad example of tight vs. tied. This is not serious editing. kwami (talk) 01:51, 15 August 2009 (UTC)

I explained this choice at the beginning of this section, and again in my response to you above. You have not answered, or even acknowledged, my reasoning. Tyuia (talk) 02:17, 15 August 2009 (UTC)

tl;dr; but I'd argue [ɹʷaɪɾɚ] ~ [ɹʷʌɪɾɚ] has as underlying repr. /raɪt.ər/ ~ /rʌɪd.ər/, so that's no more of a minimal pair between aɪ and ʌɪ than, say, Japanese /suki/ ~ /siki/ surfacing as [sˑkʲi] ~ [ɕˑkʲi] (or something like that) is a minimal pair between s and ɕ. There already is a sentence about allophones, and an example involving /l/; no point in covering all allophones of all phonemes. (If Scottish English distinguishes tide and tied, that's another kettle of fish; but that sentence about neglected distinctions isn't supposed to be a exhaustive list; there are examples such as deːz ~ deɪz for daze and days, etc. etc. etc.) -- A. di M. 20:28, 15 August 2009 (UTC)

Most of these points are covered above. The "underlying" argument doesn't work too well, as you can produce far more pairs such as spider/rider, idle/idol, eyeful/Eiffel, which can't be waved away. The only argument left is that the list doesn't need to be exhaustive, but this difference is an important one affecting Scotland and much of North America, whereas holdouts from the "long mid mergers" are few and far between. Tyuia (talk) 05:03, 27 August 2009 (UTC)

Canadian raising is common, a phonemic distinction between /aɪ/ and /ʌɪ/ is not. I think you're making the error of assuming any dialect with [ʌɪ] or some equivalent must have it as a phoneme when you've produced nothing that says as much and most literature that talks about Canadian raising talks about it as allophony rather than a phonemic contrast. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 05:11, 27 August 2009 (UTC)

I did not "assume" any dialect with that sound makes a distinction. That would be stupid. I argued that some do, and enough to make it an okay example. Tyuia (talk) 12:05, 27 August 2009 (UTC)

You cannot, in the same breath, say that this phonemic distinction is present in Scotland (which it's not) and much of North America (also not true) and that "some" dialects make this distinction. It's rare and IMHO not worth mentioning. Nothing you've presented says otherwise.

Okay, you can technically say those things (who am I to stifle free speech even amongst the incorrect?) but you're not convincing anyone. Is it really worth the fuss over mentioning a simple contrast that most people aren't even aware of? — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 20:30, 27 August 2009 (UTC)

I at least wouldn't have much problem with adding this to the key, Tyuia, if you would

a) demonstrate that the distinction applies or is neutralized equally across all major dialects, so that a transcription of /ˈspʌɪdər/ (or any other [ʌɪ] word) would work as well for RP, NZ, and Oz as it does for GA and Canadian, and

b) go through the 8450 articles linked to this IPA key and make the appropriate changes, because if the actual articles have /aɪ/, it's inaccurate of us to claim here that they are /ʌɪ/.

But meanwhile, your continued insistence on using non-examples like spite-spied to illustrate the distinction is simply perverse.

kwami (talk) 06:15, 27 August 2009 (UTC)

Again, you completely fail to grasp what is actually being said. It is listed as an example of a distinction which is not covered. Hence, it does not follow that all the linked articles must be change to cover this distinction. What you are saying makes no sense. In fact, hitherto, almost nothing you've said has made sense. Perhaps it's time to go full circle to claiming we already make the distinction? "Non-example"? Heh. Tyuia (talk) 12:05, 27 August 2009 (UTC)

On a barely related note, we don't do (a) for other splits we do cover. We don't write /ɒː/ or whatever for CLOTH words; rather, we would presumably give two transcriptions. Also, the fact that, e.g., dog is "broadened" in some LOT-CLOTH split dialects and not in others, does not render this not a real distinction. Tyuia (talk) 12:21, 27 August 2009 (UTC)

That's because we list both pronunciations in such cases.

We, all of us, grasp what you're saying. You, however, seem incapable of grasping what anyone else is saying, such as giving an illustrative example. A. di M. finally had to do that simple task for you. kwami (talk) 12:27, 27 August 2009 (UTC)

You continue to claim there is something wrong with my example when there is not, and you have failed to provide argument for your position. AM's example is worse, for reasons I already outlined several times over.

You are basically declaring separation from all use of reason by insisting that any distinction we say that we don't cover must be covered, and never admitting you're wrong to have said such things. And I'm just talking to you about the failing to grasp. Tyuia (talk) 12:54, 27 August 2009 (UTC)

How is my example worse? In most dialects with Canadian raising, /aɪ/ is only raised in syllables ending with a fortis consonant, even if "fortis/lenis" (a phonological concept) doesn't always exactly mean "voiced/voiceless" (an articulatory concept). So, in those dialects, that between /ʌɪ/ and /aɪ/ is an allophonic variation, like that between the l in pal and that in lie. (Explaining why "foolest" ~ "foo list" isn't a minimal pair is left as an exercise for the reader.) To show that they are distinct phonemes, you must show them both in the same phonological environment, as I did. --___A. di M. 14:22, 27 August 2009 (UTC)

As I've already explained so many times it's becoming tiresome, this is a page to help readers of Wikipedia. The point of this page is not to prove that these are distinct phonemes, but to illustrate what the symbols mean. To describe the split, it is sufficient to give an example of two words in which the vowel (for "raisers") is different. If I say that /ɪ/ is the vowel in kit, and /ɛ/ is the vowel in dress, I have not shown that they are distinct phonemes (the phonological environment is not identical), but I have nevertheless communicated to the reader what I am talking about.

Your example is worse for a reason I mentioned repeatedly, a point disregarded by Kwami amid his endless disingenuous insults. My example was constructed to cover two types of PRICE splitting, that of Scotland and "Canadian". Your example covers only the latter, and I'm not even sure it covers it as fully (the pronunciation of spider is less consistent among "raisers"). The text right now falsely claims this distinction is made in Scotland. But Canadian raising distinguishes (the vowels of-- duh!) tight and tide, but not tide and tied, whereas the Scottish dialects distinguish the latter pair but not the former. My example thus cleverly covers both.

This is roughly the fourth time I've explained both these points, starting with the very beginning of this section. If reason, discussion, and reaching consensus are unimportant to my fellow editors, what should I do? Just edit war? Tyuia (talk) 09:19, 31 August 2009 (UTC)

Perhaps you could take this information to the relevant articles? This is just a general information page, and the section in question is just giving examples to illustrate a point, so there's no need to add your particular example, particularly if people find it controversial.--Kotniski (talk) 09:47, 31 August 2009 (UTC)

I don't see any genuine controversy, just stubbornness. Tyuia (talk) 15:39, 31 August 2009 (UTC)

Gee, if you're not capable of conveying your point here, what makes you think you're conveying it in the key? Most readers will be utterly confused by what you propose. And you're only finding this tiresome now? Since you don't like what we now have, the solution is not to edit war, which will only get you blocked, but to delete the line, which I'll do now, and if you are able to think up an even cleverer example that actually conveys your point, we can add that at a later date. kwami (talk) 09:55, 31 August 2009 (UTC)

Please see WP:POINT.

Gee, maybe other users will read what the text says, and the fault is yours for not having done so. Tyuia (talk) 15:39, 31 August 2009 (UTC)

Hey, you were able to find an example that didn't violate the objections of every other editor here after all! and it only took three pages of petty bickering to do so. Was that so hard? kwami (talk) 18:36, 31 August 2009 (UTC)

You chose to bicker pettily, and you continue to. Although you're claiming an army of support, editors' "objections" varied. Yours was that we already cover the distinction in the key (which we never did), and that if the text says we don't cover it, then we have to change all articles to cover it. In other words, you never said anything sensible whatsoever, making it rather hard to guess what would please you. Others did make coherent points, which I replied to. I suppose the false claim that the purpose of the help text is to prove the phonemes are distinct rather than merely illustrate them is at least coherent, if silly. Tyuia (talk) 20:50, 31 August 2009 (UTC)

Kosciuszko

I need a precise IPA transcription of the local pronunciation of these American places, Kosciusko, Mississippi and Kosciuszko Bridge (New York City). In Polish, Kościuszko should be Polish pronunciation: [[] Audio file "o" not found.--Carnby (talk) 00:06, 24 August 2009 (UTC)

"kahs-kee-OOSH-koh, and most Brooklynites render it kahs-kee-OSS-ko"

I'm assuming there's no difference in the vowels of kahs and oss? If so, these would be /kɒskiˈʊʃkoʊ/ (or perhaps /-uːʃkoʊ/) and /kɒskiˈɒskoʊ/, with only the stressed vowel differing from the Aussie mtn.

"Kah-zee-ESS-ko"

They presumably mean /kɒziˈɛskoʊ/.

kwami (talk) 00:47, 24 August 2009 (UTC)

There's also the possibility of reduction of the first unstressed syllable. Thus /kəs-/ in all cases, right?--Carnby (talk) 07:50, 24 August 2009 (UTC)

I wouldn't know. Do you know that the first syllable is reduced? kwami (talk) 09:44, 24 August 2009 (UTC)

To be honest, no. But it is reduced in many English words, due to the peculiar prosody of the language.--Carnby (talk) 11:19, 24 August 2009 (UTC)

But there are plenty of exceptions to that, and our only evidence suggests that this is such an exception. We'd have to attest that it's /kəs-/ before we make any such claim. kwami (talk) 06:19, 27 August 2009 (UTC)

The Australian mountain is pronounced as either /kɒziːˈɒskoʊ/ or /kɒˈʃʊʃkoʊ/ ("ko-zee-OSS-koh" or "ko-SHOOSH-koh" respectively).

Have you consulted a dictionary? – Marco 79 15:15, 27 August 2009 (UTC)

LPD by J.C. Wells says /kɒsiˈʌskəʊ, -ˈʊsk-/ for British English and /ˌkɑːs- ˌkɑːsk-/ (with /-oʊ/) for American English. So I think the Brooklyn bridge could be /ˌkɑːskiˈʊʃkoʊ/ or /ˌkɑːskiˈɒskoʊ/ (/-ˈʌ-/?). The place in Mississippi could be /ˌkɑːziˈɛskoʊ/ as in the US kah- could stand for /ɑː/--Carnby (talk) 22:56, 31 August 2009 (UTC)

In the US, kah- would also stand for /ɒ/, so we might as well use that. The question is which place Wells was referring to. None of his variants fit the descriptions we have of either Brooklyn or Mississippi. Best bet might be to call up s.o. in the area, like the local library, and ask them directly. I might get around to that tomorrow. kwami (talk) 00:24, 1 September 2009 (UTC)

IPA key display

I know we've gone over this before, but maybe browsers have evolved a bit. Does this work for people?

The last sample word should be hew. (I'm not proposing to remove the link, I just haven't bothered to add it here.) kwami (talk) 00:29, 7 September 2009 (UTC)

An option with shorter strings could be

though I don't know how we communicate that it's there. kwami (talk) 01:35, 7 September 2009 (UTC)

It works for me. Using it to be redundant (that is, saying it's pronunciation in the IPA) shouldn't be a problem since someone who doesn't hover over the text wouldn't miss too much. What else would we use it for? — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 20:06, 7 September 2009 (UTC)

{{pron-en|fuː}}^key. --___A. di M. 22:27, 7 September 2009 (UTC) (The capital I in "hit" looks too much like a small ell, could we use "bUsiness" instead? --___A. di M. 22:30, 7 September 2009 (UTC))

The reason for using the shortest possible words is that many browsers clip the display. That's why I have the alt. w the key broken into three parts. Also, w the minimal set, there isn't much reason to capitalize. kwami (talk) 22:41, 7 September 2009 (UTC)

We can't have a minimal set for consonants as /j/ and /ŋ/ are in complementary distribution, and if we capitalize consonants we can capitalize vowels as well for consistency. (Unless we classify /j/ as a "semivowel" and have "bathe, badge, bang, bath, bash, batch, beige"; that doesn't work for those who have the FATHER vowel in BATH, but I guess they'd get it anyway; and "bathe" and "beige" won't fit, either.

key --___A. di M. 23:14, 7 September 2009 (U

Comments On The Vowels Of Wikipedia’s IPA For English

The articles in question are Wikipedia: IPA for English and International Phonetic Alphabet, which I will call IPAE and IPA:

<a href="http://en.wikipedia.org/wiki/IPA_for_English">Wikipedia: IPA for English</a>

<a href="http://en.wikipedia.org/wiki/IPA">International Phonetic Alphabet</a>

1.) Comparing IPAE with the vowel trapezoid shown in IPA, one would conclude that the vowels of code and fade are diphthongs, but in General American English they are monophthongs. Grammarians and lexicographers merely copy each other when they almost invariably call them diphthongal.

2.) One would also conclude that the vowels of food and bawd, as well as the first vowel of the supposed diphthong in code are rounded vowels, each having an unrounded counterpart found in some other language. This is not generally true. The vowel of good also is treated as a rounded vowel with no attested unrounded equivalent. However, there are no rounded vowels in General American English. Rather, the vowels of food and code are usually unrounded vowels having as rounded counterparts vowels not unlike the ‘u’ and ‘o’ of Spanish. Rounded counterparts of the vowels of bawd and good can be produced readily, even if their occurrence in natural languages is not attested.

3.) Furtive shwas do not occur before the ‘r’ of such words as bear, beer, boar and boor, as shown in IPAE. True enough the vowels are r-colored but free of glides. Furtive shwas appear before ‘r’ after true diphthongs, as in fire and hour, and rarely in words like seer, when it is the agent of see.

4.) The vowels of bawd and bud do not constitute a rounded-unrounded pair, as one would gather by comparing IPAE and IPA. Both are unrounded, with the former articulated just below the latter, in the back of the mouth. Either may generate a rounded counterpart, but I don’t know which languages, if any, have such counterparts.

5.) The vowels of balm and bot are identical in General American English, as noted in Note 8 in IPAE. However they are not back vowels. They are front vowels, articulated just below the vowel of bat. There is no back vowel at this height in General American English.

6.) Although Wikipedia passed over this point in silence in their article IPAE, I assume that they would consider the vowels of beg and bed to be identical, but the vowel of beg is the same vowel as the monophthong of bay. This happens only before ‘g’ and ‘ng’.

7.) A furtive shwa appears before ‘l’ in words like feel, file, fail, foil and mule. It sometimes is heard in words like vowel, where it can be said to be represented by ‘e’, but more often vowel is pronounced like Val, with an l-colored version of the vowel of bat.

8.) The vowel of bag, bang, bank and bash is not the vowel of bat, but passes over into a falling diphthong, with furtive ‘i’ following. This is limited to the consonants specified.

9.) Similarly, the vowel of wash sometimes tends to diphthongization, with furtive ‘i’ following. This occurs only with ‘sh’.

10.) I note that Wikipedia’s IPAE does not treat the final syllable of each of the words button, rhythm, bottle and mercer as a syllabic consonant, as some dictionaries do. I agree with IPAE on thus point.

Thomas Keyes

I've fixed the header: please use "equals" signs or the "new section" link at the top of page, so that the discussion shows up in the TOC, we get the "edit" link for the section, etc. BTW, this is a phonemic transcription, so irrelevant differences such as [o] in AmE boat vs [əʊ] in BrE boat, or [l] in line and [əɫ] in feel aren't transcribed, but we use the semi-arbitrary phonemic symbols /oʊ/ and /l/. (Now, there are dialects with a monophthong in toe and a diphthong in tow, but we aren't even attempting to transcribe all distinctions in all dialects.) As for syllabic consonants, that depends very much on the speaker; but I wouldn't surprised if there were someone using /sɛvn/ for seven and /sɛvən/ for Severn, and saying that their accent has a phonemic schwa in seven would be unambiguously false. --___A. di M. 12:50, 12 September 2009 (UTC)

It wouldn't appear that you speak GA. /e:/ and /o:/, for example, are obvious diphthongs in GA, and /a:/ is clearly a back vowel. But that's irrelevant, as this is not an IPA key for GA. kwami (talk) 18:13, 12 September 2009 (UTC)

I think the confusion here is that User:Thomas Keyes is looking at WP:IPA and noting phonetic discrepancies. There are such discrepancies, but this is a key that is designed to accomodate multiple dialects so, for example, if your dialect doesn't have a pre-rhotic schwa in beer then you can ignore it. — Æµ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:52, 12 September 2009 (UTC)

BATH vowel

In Southern England the vowel of bath and most /æ/ before fricatives is the same vowel as father. What would the compromise IPA transcription be for this BATH vowel? I suspect and would favor /æ/ as it's more common and more conservative. This is relevant for English words which may need to be transcribed for clarification or demonstration and for foreign words with the Israeli city Jaffa being a possibly affected. I was prompted to ask this by the transciption of photograph as /ˈfoʊtəɡrɑːf/ in the English orthography article. Jackessler (talk) 00:11, 30 September 2009 (UTC)

Well, we have a few choices. We can transcribe such words twice, once for each vowel sound; we can choose one over the other, as you suggest; or we can introduce a new letter—<a> has been suggested—to indicate a vowel that varies per register. This hasn't been discussed at any length. If the variation is largely predictable, the way that /j/ dropping is predictable in GA, then we should go with your suggestion; exceptions can be marked by double transcribing. If it is not all that predictable, then we can go one of the other two routes; objections might be raised about introducing a new distinction to remember, and <a> suffers the problem that it is often used to mean specifically /æ/ or specifically /ɑː/, though I think I've managed to eliminate most cases of that which transclude an IPA template. kwami (talk) 00:29, 30 September 2009 (UTC)

It's not predictable; see Variations, and there are minimal pairs such as aunt vs. ant. As for choosing one, many dictionaries pick /ɑː/ because that's what RP does, or give /ɑː/ as default and /æ/ as "American" (despite the fact that the latter is widespread everywhere except southern England, New England, and the southern hemisphere); I don't think this is a good idea. We're left with either listing both pronunciations (which makes perfectly sense, given that for many speakers they're in free variation), or using another symbol such as /a/ (it'd make sense if the claim that some ‘use in "ask" a vowel distinct both from the one in "cat" and the one in "father"’ is true; of course that's an awful example because most people with such a three-way distinction are those with phonemic æ tensing /eəks/, but if there are non-rhotic speakers with a three-way ant–aunt–aren't distinction (or father–bother mergers with three-way ant–aunt–font) then /a/ would be the only sensible way to show the latter; but I suspect these dialects are very rare). Anyway, I've no strong opinion: either a double transcription or an extra symbol would work for me. (Ditto for CLOTH.) ___A. di M. 22:08, 30 September 2009 (UTC)

Yes, I was referring to your earlier suggestion. Where we make the distinction, we currently do it with double transcription; personally, I really like your idea of having a dedicated letter for this, but before we introduce such a distinction, I think we should get some general agreement that it's a good idea. We have introduced new distinctions before, but only with reduced vowels, which people aren't as sensitive to.

How common lexically is the CLOTH distinction? The obvious solution would be to use one of the existing symbols and change the length sign; however, leaving it off /ɔː/ is a frequent variant already, so it might be best to add length to /ɒ/ for "/ɒː/". I don't know if that would go over very well, though, as I've never seen a dictionary that uses it, and people might object about OR. kwami (talk) 22:35, 30 September 2009 (UTC)

I've seen no dictionary using ɵ, either. Also, I'd prefer /ɔ/ for consistency with the fact that other checked vowels are transcribed with one character and free ones with two; unfortunately it clashes with /ɔr/. Maybe we could continue with inventions and use /ɔ/. :-) (No, I'm not serious.) In any event, my proposal was more a "what if we did this?" than a serious one; having double transcriptions for BATH and CLOTH is good enough.

As for your comment that /a/ is unpredictable, is it at least predictable before frics, as in -graph? kwami (talk) 00:08, 1 October 2009 (UTC)

In RP, class doesn't rhyme with mass, gaff doesn't rhyme with half, halve is different from have, castle doesn't rhyme with hassle, yadda yadda yadda. (It looks like they're nearly equally split before /s/, /ɑː/ is more common before other voiceless fricatives, and /æ/ before voiced ones.) OTOH, I think it's unlikely anyone pronounces some -graph words with PALM and other -graph words with TRAP consistently. ___A. di M. 00:51, 1 October 2009 (UTC)