Help talk:IPA for English

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Inconsistent redirects[edit]

Hi there, I find it a bit silly that one lands in a different page depending on whether IPA is spelled all uppercase or not... Compare:

I would say we should make it consistent (my preference is for always redirecting to article space). Thanks. (talk) 06:28, 23 July 2016 (UTC)


I, & many other speakers of accents similar to IPA, pronounce situation as [ˌsɪtʃuˈeɪʃn̩] or [ˌsɪtjuˈeɪʃn̩]. The u would be considered to be /juː/. I have never heard situation pronounced [ˌsɪtʊˈeɪʃən] by a native English speaker. Dr. British12 (talk) 18:42, 30 August 2016 (UTC)

+1 Peter coxhead (talk) 21:15, 30 August 2016 (UTC)

North–force distinction[edit]

Help:IPA for English § Dialect variation states that the Wikipedia diaphonemic system accommodates General American, Received Pronunciation, Canadian English, South African, Australian, and New Zealand. That is accurate, but that is not all that is being accommodated, because the diaphonemic system distinguishes between /ɔːr/ and /ɔər/ (lexical sets north and force). According to the vowel table at International Phonetic Alphabet chart for English dialects, none of the above-mentioned dialects make this distinction. Three in the table do: Irish, Scottish, and Welsh English (what you might call Celtic-influenced varieties). If I recall right, some varieties of New England English and Southern American English also make the distinction, but they are not in the table.

So, which of these varieties of English is the inclusion of the north–force distinction meant to accommodate?

Scottish English and some varieties of Irish English have the distinction, but they also make the distinction between fern, fir, and fur, so it cannot be for their sake that the distinction is included, or else the distinction between fern, fir, and fur would also be included. Whichever dialect is actually being accommodated by the north–force distinction should be listed, for the sake of completeness.

And why does the diaphonemic system accommodate this particular distinction and not the fern, fir, fur one? — Eru·tuon 00:24, 2 September 2016 (UTC)

This distinction is possible in General American (as [ɔɻ] vs [oɻ]), but is probably in rapid decline. Mr KEBAB (talk) 00:25, 2 September 2016 (UTC)
I have never heard it, so I assume it is also geographically limited. I realized I should look at Rick Aschmann's site, and he lists Eastern New England, Birmingham (Alabama), and New Orleans. He doesn't mention whether it is in decline there or not. — Eru·tuon 01:20, 2 September 2016 (UTC)
According to Wells, in addition to Scottish and Irish English, dialects in the American Northeast also make said distinction. The important thing here is what is found in dictionaries. If we can't back up the distinction with consistent sourcing, then we can't encode it in our system. In this case, we have an artifact of an earlier distinction in RP that RP no longer has, can be found in older/traditional dictionaries, and is also still found in various dialects. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 02:35, 2 September 2016 (UTC)
If sources are such a problem then, by all means, let's stop making that distinction. I don't think anyone would really care. Mr KEBAB (talk) 03:18, 2 September 2016 (UTC)
Are sources a problem? I thought they weren't. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 14:42, 2 September 2016 (UTC)
I don't know. I said 'if'. We need to check this. Mr KEBAB (talk) 14:45, 2 September 2016 (UTC)
This should clarify the matter. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 17:47, 2 September 2016 (UTC)
Thanks. So, according to that article, only and an unknown number of dictionaries using the Americanist phonetic notation make that distinction. Longman Pronunciation Dictionary (but not CEPD) also makes that distinction. Mr KEBAB (talk) 17:58, 2 September 2016 (UTC)
@Mr KEBAB: I know that Tharthan speaks a New England dialect with a partial lack of the horsehoarse merger and he might object to the distinction being removed. But the symbol for hoarse does not really describe the actual pronunciation in his dialect. Otherwise, you are probably correct. — Eru·tuon 19:11, 2 September 2016 (UTC)
Indeed, thanks for pinging me, Erutuon. I have a lack of the horse-hoarse merger in words such as "hoarse", "mourn", "four", "coarse", etc. I would object to removing the distinction here, as such. Tharthan (talk) 20:18, 2 September 2016 (UTC)
@Erutuon and Tharthan: We have no reason to remove the distinction as there are enough sources to back up transcribing it. Plus, it's an optional part of General American (see e.g. the LPD). Mr KEBAB (talk) 04:44, 3 September 2016 (UTC)
I (theoretically) speak a New England dialect (though notably NYC-influenced), and yet I certainly make no distinction between, for example, horse and hoarse, nor does anyone I've heard in the various Western New England towns where I've studied and lived. Regardless, however, I agree that the sources should be our basis for decisions on this matter. Wolfdog (talk) 20:35, 2 October 2016 (UTC)
@Ƶ§œš¹: So, there are no dictionaries for Scottish English that would distinguish between the vowels of fern, fur, and fir? — Eru·tuon 19:05, 2 September 2016 (UTC)


Currently, "oral" is listed as an example of a /ɔər/ word. As someone with an Australian accent, while I would pronounce the vowel in the other examples listed ("force", "more", and "boar") identically, I would pronounce "oral" such that it rhymes with "moral" (which is listed as /ɒr/). As such, since this page is supposed to include Australian pronunciations, I would recommend removing the example altogether. There is already an example of "or" yielding that vowel sound, so I don't think a replacement would be necessary. If one is, I would suggest "aural" (which I would pronounce the same way as the other examples, and I'm led to believe it is a homophone for "oral" in other relevant accents). --SnorlaxMonster 06:37, 24 September 2016 (UTC)

Retracting that last suggestion. According to Lexical set, "aural" is an example of the set /ɔːr/ ("north", "born", "war", "Laura"). Missed that, because I have the Horse–hoarse merger. --SnorlaxMonster 07:16, 24 September 2016 (UTC)

Marking transcriptions explicitly as diaphonemic[edit]

It's very misleading that this system uses slashes / /, when those are generally used for phonemic transcriptions. I have seen many people who have been misled into thinking that this transcription system is phonemic or phonological, when it is not. Most recently, Jérôme proposed that Wiktionary create a phonemic transcription system like that used on Wikipedia, when what he was really proposing was a diaphonemic system. Ardalazzagal stated on the talk page of Theresa May that John has the same phonemes in all dialects of English: /ɒn/. That's wrong: it has the phonemes /dʒ/, /ɔ/, /n/ in modern RP, AuE, and NZE, but the phonemes /dʒ/, /ɑ/, /n/ in GA and /dʒ/, /ɒ/, /n/ in Canadian. The consonant phonemes directly correspond; the vowel phoneme does not, because there are different numbers of vowels in each dialect, with different phonological features distinguishing them from each other. The vowel phonemes do generally correspond in the same words, but that doesn't make them the same thing.

I defended the Wikipedia diaphonemic system before, but I think it really needs to be distinguished visually from phonemic transcriptions. In a previous discussion, someone mentioned double slashes // //, pipes | |, or braces { } as possibilities. Pipes and braces would difficult or confusing to use inside of the {{IPA}} template, but they could be generated by {{IPAc-en}}. Whichever option is chosen, it would be better than the current untenable situation, in which people are being regularly misled that the Wikipedia system is somehow phonemic rather than diaphonemic and that it indicates that the various English dialects have the same vowel system (not true!). — Eru·tuon 17:11, 14 October 2016 (UTC)

The most important thing to determine here is whether people, in confusing the diaphonemic system for a phonemic one, mispronounce the transcriptions. In your two examples, you seem to be fixating on a terminological inaccuracy rather than a practical one. In the case of Jérôme, he explicitly described our system in a fairly accurate way and was simply mislabeling it. This is understandable, considering the distinction between phonemic and diaphonemic is made for and by specialists.
So let's not get ahead of ourselves without real evidence. I would hate for this to turn into a more academic dispute, again, with no real evidence that users are actually confused by our stylistic choices in a meaningful way. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 00:38, 15 October 2016 (UTC)
Well, there was the case where an Australian (Peter Greenwell) said that Australian DRESS couldn't be pronounced /e/, because FACE is pronounced /eɪ/. He was being misled by the transcription // into thinking that the vowel in FACE was /eɪ/ in Australian, whereas it is actually /æe/. To be fair, I'm not sure if marking the transcriptions would prevent this type of confusion. But at least marking diaphonemic and phonemic transcriptions differently would encourage people to learn the distinction between diaphoneme and phoneme, rather than encouraging them to assume there is none. You can see how confusing it is here, because I used the same markings for the diaphoneme // as I did for Australian English phoneme /æe/! It almost sounds like a contradiction. It would make more sense if I used //eɪ// and /æe/. Then you can see that the two symbols are different categories of thing. — Eru·tuon 01:16, 15 October 2016 (UTC)
Whether Australian FACE is transcribed /æɪ/ or /eɪ/ is a matter of a rather arbitrary choice. You may want to read [1], specifically this: "Don't ever ask us which of these systems is the IPA. They both use IPA symbols, but they are both local conventions used for phonemic transcriptions of Australian English. The Mitchell-Delbridge system is quite similar (but not identical) for a system used to do phonemic transcriptions of the British Received Pronunciation (RP) dialect, whilst the Harrington et al system more closely represents the actual pronunciation of an average modern speaker of Australian English." I'm afraid you're just being pedantic here, as phonemic symbols do not have to match the most common phonetic realization (although, of course, it is nice and useful when they do, but let's keep it real, as this isn't about our wishes.) By the way, [eɪ] (or [ɛɪ]) also occurs in Australia. I've been listening to AuE long enough to say that it's a fairly common pronunciation which isn't found only in the Cultivated variety (as one table on Australian English phonology states), but the General one as well. I wouldn't expect it to be common in the Broad variety, which has a very Cockney-like diphthongs. Mr KEBAB (talk) 02:08, 15 October 2016 (UTC)
@Mr KEBAB: Very well, /eɪ/ is a legitimate phonemic transcription of Australian, but it seems to have caused a misunderstanding in the case of the editor I mentioned. You're not responding to the fact that, if phonemic transcriptions actually differ from diaphonemic transcriptions, our system does not distinguish the two visually, which is likely to cause confusion and to discourage people from making a proper distinction between the two. — Eru·tuon 02:17, 15 October 2016 (UTC)
Well, all I can do is agree. Mr KEBAB (talk) 02:25, 15 October 2016 (UTC)
I vaguely recall this being brought up before, but I don't remember the rationale for going against changing the slashes. Perhaps someone can scour the archives more closely than I have. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 04:55, 15 October 2016 (UTC)
@Aeusoes1: I remember bracketing being briefly suggested when the then-new (eventually rejected) symbols //ɒː// and //aː// were being discussed. I think it wasn't the main topic of conversation, so nothing happened either for or against it. — Eru·tuon 05:10, 15 October 2016 (UTC)
I fully agree that we should mark our diaphonemic transcriptions as what they are: diaphonemic. In not doing so, we are misleading linguistically inclined readers (anybody who has ever learned about the /…/ convention for marking phonemes) into assuming that we use phonemic transcriptions, and we are misleading linguistically trained editors into using phonemic transcriptions on our articles. --mach 🙈🙉🙊 07:19, 15 October 2016 (UTC)

So, Mr KEBAB and J. 'mach' wust have expressed support for the idea of visually distinguishing diaphonemic transcriptions from phonemic. I'm curious, do you two, or anyone else, have a preference for what markers should be used?

As I mentioned before, I think pipes, curly brackets, and double slashes are options: |ðɪs|, {ðɪs}, or //ðɪs//. I have no idea if these symbols are actually used to mark diaphonemic transcriptions in sources, but I vaguely remember them being mentioned in an earlier discussion (which I can't find for some reason...). Curly brackets and pipes would both be hard to type in wikitext (and might be mistakenly "corrected" by random editors), so I'm leaning towards double slashes.

Someone might think that {{IPA|{ðɪs}}} (though {{IPAc-en}} should be used instead) was an attempt to insert the syntax for a template parameter – {{{1}}} – or that the extra bracket indicated a syntax error. And vertical bars (pipes) have to be typed using {{!}} or |, both of which are more laborious than //. — Eru·tuon 23:39, 20 October 2016 (UTC)

Let's use double slashes. Mr KEBAB (talk) 00:31, 21 October 2016 (UTC)
Probably best not to go ahead with the change yet. I posted on the Linguistics and Languages WikiProjects to see if I can get more than four people to post in this thread. — Eru·tuon 02:14, 21 October 2016 (UTC)
You asked me for my preference, so I answered. Mr KEBAB (talk) 03:30, 21 October 2016 (UTC)
The article on diaphonemes mentions a few representations at the end. Some of them, such as pipes, have been used to represent something else, such as morphophonemic or morphemic representations. If we're going to do something different than slashes, double slashes seems the easiest solution. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 04:18, 22 October 2016 (UTC)
In my opinion this discussion is of largely theoretical interest, without merit for the purpose of this template and the understanding of our readers. Yes, a diaphonemic transcription system is not the same as a phonemic one, but when any of these transcriptions is offered, together with a key of how to pronounce each (dia)phoneme that is sufficient information for the users to get a good impression of the real pronunciation in his favourite dialect. Therefore I see no need for an explicit marking and would prefer to avoid the additional clutter it would create. −Woodstone (talk) 08:40, 21 October 2016 (UTC)
@Woodstone: The additional clutter is minimal: only two extra characters in the displayed text – //ðɪs// versus /ðɪs/. As far as it being theoretical, so is the distinction between phone and phoneme, yet we attempt to consistently mark it throughout Wikipedia. I am proposing that we do the same with diaphonemes. Theory has practical meaning in this case: diaphonemes are farther removed from the actual default phonetic value than phonemes are, and so looking at the diaphonemic symbols and expecting that they indicate something about the default phonetic value of those symbols will lead people even farther astray. — Eru·tuon 17:49, 22 October 2016 (UTC)

@Erutuon: (Sorry for such a long text. I did not intend to write such a big piece, but that's how it came out.) I suggest we may go to the opposite direction. That is to ask people not to understand slashes for what they are not supposed to represent. Many think that the narrow transcription with square brackets must be really very narrow and represent an actual realisation at a given moment or a situation with minute details, thus they try to avoid it altogether and use the phonemic notation with slashes. But this generates more and more "phonemes" which, in fact, hardly exist. So when people write things like /æe/ for AuE or whatever, it does not mean there is an independent AuE phoneme in FACE entirely different from the FACE phoneme in other varieties of English, it actually implies that the general English phoneme (and calling it a diaphoneme is redundant here) /eɪ/ is realised in AuE as [æe], even if [æe] is not the only realisation (but at least very widespread), and even if further transcription details might be added. So when some people say, as you reported, Australian DRESS couldn't be pronounced /e/, because FACE is pronounced /eɪ/, it doesn't make sense. Neither /e/ nor /eɪ/ are the actual representations of pronunciation or that it have to be pronounced exactly that way as written, but abstract representations of the English phonemes (again it is not always necessary call them diaphonemes). We might use an entirely different notation and symbols for them, for example, the Shavian 𐑧 and 𐑱 (I hope you have a proper font to see them). Imagine the very same sentence with these symbols: Australian DRESS couldn't be pronounced 𐑧, because FACE is pronounced 𐑱. How does it sound? Indeed, it entirely does not make any sense.
The problem, as I see it, is not about the slashes, but that people do not like the fact that the widespread symbols for the English phonemes have been designed in accordance with RP, where these symbols are not only abstract representations, but more or less represent the actual pronunciation. This is why people in Australia, America or wherever try to invent their own "right" IPA notation for the very same English phonemes with the very same distribution. (The examples are the Australian and Canadian versions of Oxford dictionaries where their localised transcription hardly add any useful information, apart from that Aussies realise vowels differently than Brits, or Canadians round their vowel in LOT.) But if somewhere it might be justified on various grounds, here there must be no place for national pride. Even if people may not like that the transcription for all varieties of English is set up by those British posh academics according to how they or their circles pronounce English. Just a personal side-note. I always favoured the IPA for English and thought it is the only proper way to represent English pronunciation, but the more I study English accents and varieties, the more I understand Americans who do not like the IPA at all. Not only their dictionaries, but American linguists as well. Because even if that RP-centric IPA for English is good from a pedagogic point of view and may teach a "correct" pronunciation (though I don't think at all RP is the best for ESL learners), it is very, not even British-centric, but Oxbridge-posh-centric. For ESL learners it may be not an issue, but for native speakers of English it is a problem when they see /eɪ/, but know, especially if they know the entire IPA as well, that they do not pronounced that way.
Returning to our main question. I think the transcription used in Wikipedia already represents it right. It does not supposed to represent the actual pronunciation. So a good solution may be to leave the things as they are, but just to explain people that symbols between single slashes is not the way they pronounce or ought to pronounce, these are just abstract symbols which merely give a hint about pronunciation. This is what phonemes actually mean. No need to invent diaphonemes and signify them in a special way.
And another personal opinion. This misunderstanding is why I now somewhat prefer respelling than the IPA outside of the field of phonetics. The IPA is good for scholarly and academic writings, but not for laymen with multitude of accents. The "IPA of dictionaries" is too skewed toward one variety. But this skew is inevitable if we want a well-understood by everybody uniformity. Thus here may be another solution: not to use the IPA, but the respelling which everybody may interpret as they want. But I understand this is not going to happen as the IPA is very well established in Wikipedia. So let's leave those single slashes as they are. They already signify what they are supposed to. But if youse agree upon double slashes, I'll not oppose it, but I think it will cause no less confusion. Many people will continue to use single slashes anyway just due to the habit.--Lüboslóv Yęzýkin (talk) 21:07, 22 October 2016 (UTC)

As long as we are using diaphonemic transcriptions, I think it is very important that we improve our article about diaphonemes. When a reader follows the link on our transcriptions, it will ultimately lead them to that article. At the moment, the reader could get the mistaken impression that the purpose of diaphonemes is having a single symbol for a range of allophones. However, that is rather what the purpose of phonemes is, not of diaphonemes. --mach 🙈🙉🙊 16:53, 27 October 2016 (UTC)
It might be best to have a discussion on improving that article at Talk:Diaphoneme, rather than here. I agree that a good diaphoneme article goes hand-in-hand with using a diaphonemic transcription. But I'm not sure I see how that confusion can take place. Could you be more explicit? — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 19:32, 27 October 2016 (UTC)
See Talk:Diaphoneme#Lead rewrite. --mach 🙈🙉🙊 07:18, 29 October 2016 (UTC)
Oh yes, I remember that thread. You said you were going to replace an example in the lede. If that's the only thing wrong with the diaphoneme article, why are you bringing it up again rather than just fixing the problem you had with it? — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 20:40, 29 October 2016 (UTC)
I didn't answer back then in October/November just to give myself a little time to think about the issue and nearly forgot about the discussion, but it's better late than never.
I must beg your pardon if I'm not too skilled in theoretical lingusitics as many might be here, as my interest in languages has been always leant toward more trivial and practical fields than to highly abstract theoretical ones. Probably in my magnum opus from the above I wrote some naive things or misconceptions. Frankly, I have not met the term "diaphonemes" often, but even after having read the article I still have no clear understanding what is this exactly and why this exists at all. From the "Dialectology" section I understand that from the point of view of the practical applications diaphonemes are phonemes with a broader umbrella meaning, so I do not understand why /njuː ˈjɔrk/ cannot be seen as a simply phonemic transcription with two phonetic regional "broad" realizations, [nuː ˈjɔrk] and [njuː ˈjɔːk]. Actually, postulating that English of all varieties may have the common phonemic system with—even of not always apparent but underlying—phonemes and, further, with different "broad" realizations in each variety and multitude of "narrow" idiosyncratic realizations is no less legit than postulating that there are diophonemes, several phonemic systems in each variety and multitude of phonemic realizations. Overall, diaphonemes complicate things and limit the understanding the term "phonemes" too narrowly. But I'm not going to continue as I'm not so skilled on that very abstract theoretical subject. But what is clear for me that, if I could not have got an idea, there is a guaranteed chance that our layman readers will not understand it, either. And for most this "phonemic vs. diaphonemic" scientific dispute will certainly remain simply out of their understanding. But all this obscurity results more on the fact that there seems to be no explicit theory of diaphonemes, at least that can be applicable directly to English, neither there seems to exist an established diaphonemic transcription which we could follow as an example. So what we're discussing here, and I dare to say this, is more or less an original research. Even if we change single slashes to double slashes it won't cease to be an OR, our own Wikipedia's original interpretation of diaphemes in English. Returning to our article "Diaphonemes", a diphonemic transcription of late, wait, eight must be //let//, //weit//, //ext//, but it is not at all what we are doing here, but we write (I won't use any delimiters deliberately): leɪt, weɪt, eɪt, that seems for me nothing more than a phonemic transcription. What we actually did we took the RP transcription used in dictionaries and tweaked it to cover some other accents, particularly General American, which we simply cannot ignore (but we ignore other accents easily). If I'm not wrong our system is an American-British mix of that of Cambridge's English Pronouncing Dictionary. But it does not seem for me diaphonemic. Probably we might use a truly diaphonemic transcription, but we then could not have done with just changing the number of slashes, we must rewrite our system completely.--Lüboslóv Yęzýkin (talk) 18:01, 22 January 2017 (UTC)
I think the article diasystem might help you understand some of the issues. A diaphonemic representation does not have to be panlectal. That is, it need not encompass all English dialects. Ours is what one might call polylectal, as it focuses only on a handful of standard varieties. Indeed, the characterization you make of the system we use as starting off with British English and incorporating features of American English is pretty much how it was characterized until the term diaphoneme was uncovered as an easy way of describing this.
More importantly, our system as we use it doesn't depend on people having an intimate understanding of theory behind diaphonemes or the distinction between phonemes and diaphonemes. The diaphonemic identification of different phonemic systems at the heart of it is what people do when listening to speech of other dialects. — Ƶ§œš¹ [lɛts b̥iː pʰəˈlaɪˀt] 20:07, 22 January 2017 (UTC)
@Lüboslóv Yęzýkin: Your reservations against the term diaphoneme are felicitous. Similar reasons lead linguists to drop the term decades ago. The question whether or not inventing our own transcription system constitutes original research is a contentious issue: Some editors (myself included) think that inventing our own transcription system is original research, while other editors think it is not. I think you have a point that our system is basically RP with a few tweaks. Take the OED2 transcription, add rhotic /r/, replace /əʊ/ by /oʊ/, introduce OED3 /ᵻ/ and /ᵿ/ and voilà, that is pretty much the system we are prescribing. --mach 🙈🙉🙊 20:52, 22 January 2017 (UTC)

Introduce ambiguous tooltips for /ɪər, ɑːr, ɔːr, ɜːr/ in order to transcribe words like “Nero, safari, aural, McMurray”[edit]

Please see Template talk:IPAc-en#Introduce ambiguous tooltips for /ɪər, ɑːr, ɔːr, ɜːr/ in order to transcribe words like “Nero, safari, aural, McMurray”. --mach 🙈🙉🙊 23:43, 10 January 2017 (UTC)