Archives

October 2007 to March 2008 April 2008 April to December 2008 January to February 2009 March to May 2009 June 2009 to July 2009

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)[reply]

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 09:44, 3 August 2009 (UTC)[reply]

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)[reply]

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". +Angr 10:44, 3 August 2009 (UTC)[reply]

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)[reply]

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 08:31, 5 August 2009 (UTC)[reply]

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)[reply]

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)[reply]

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. +Angr 14:42, 5 August 2009 (UTC)[reply]

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)[reply]

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. +Angr 15:16, 5 August 2009 (UTC)[reply]

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)[reply]

/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. +Angr 16:47, 5 August 2009 (UTC)[reply]

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)[reply]

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

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)[reply]

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

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)[reply]

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)[reply]

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)[reply]

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)[reply]

"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)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

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. +Angr 14:36, 6 August 2009 (UTC)[reply]

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)[reply]

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:46, 6 August 2009 (UTC)[reply]

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)[reply]

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

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

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 23:20, 6 August 2009 (UTC)[reply]

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)[reply]

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

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

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)[reply]

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

• Help talk:Pronunciation/Archive 1#IPA Pronunciation Irrelevant To Most Visitors to Wikipedia
• Help talk:Pronunciation/Archive 1#Controversial, OR
• Help talk:Pronunciation/Archive 1#annoying gibberish
• Help talk:Pronunciation/Archive 2#Is there value in this at all?
• Wikipedia talk:IPA for English/Archive 3#While Chart Is Useful - It Violates All Wikipedia Principles
• Wikipedia talk:IPA for English/Archive 3#Petition to develop a more user-friendly pronunciation standard for articles

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

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)[reply]

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 14:12, 13 August 2009 (UTC)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

'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)[reply]

"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)[reply]

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)[reply]

/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)[reply]

/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)[reply]

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

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)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

/'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)[reply]

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)[reply]

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 14:47, 13 August 2009 (UTC)[reply]

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. +Angr 15:03, 13 August 2009 (UTC)[reply]

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)[reply]

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

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)[reply]

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)[reply]

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)[reply]

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

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)[reply]

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

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)[reply]

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. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 04:12, 14 August 2009 (UTC)[reply]

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)[reply]

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)[reply]

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)[reply]

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.) +Angr 09:04, 14 August 2009 (UTC)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]

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)[reply]