Wiktionary talk:Pronunciation

Definition from Wiktionary, the free dictionary
Jump to: navigation, search
See also the Collected archive of pronunciation discussions from the community fora.

Not just English[edit]

Dear Peter Isolato, with regard to your criticism on my talk page about removing the English pronunciation key and instead setting up a "mini-portal" on this page, please note the following: Having an English pronunciation key on a page entitled "Wiktionary:Pronunciation" belies its name. There are so many more aspects of pronunciation we have to deal with. These include For users:

For editors:

Having a pronunciation portal seems a very good solution. It's incomprehensible to me why you think it's a bad move. Of course, the page needs to be expanded and have some explanatory text rather than just being a list of links. My setting-up of this portal was prompted by a string of recent posts by users who were seeking information on pronunciation and seemed lost/unable to find the relevant pages. I remember and were able to find these: Help_talk:Contents (on 17/03 by and WT:BP#Pronunciation Guide/Audio Files - A single location to discuss? (on 16/03 by Ncik 14:09, 20 March 2006 (UTC)

Pronunciation guidelines[edit]

Copied from discussion in Wiktionary:Beer parlour archive/November 06

There are extensive charts about IPA and SAMPA rerpresentation of English and other sounds, but what is the guideline about stress symbols (ˈ) and syllable breaks: (.)? Most French entries seem to have syllable breaks, but no stress symbols. English entries seem to have stress symbols but no syllable breaks. Is there a consensus about this, which should be included? henne 12:00, 23 November 2006 (UTC)

Stress is less important in French than in English: in French, syllables are generally evenly stressed, except the last syllable is stressed at the end of a group of words. I don't believe that the syllable break symbol is used much, at least not in my French-English dictionary. Poccil 22:17, 24 November 2006 (UTC)
Adding syllable breaks to English words isn't easy. Consonants at the end of a word tend to be tied into the next word, e.g. red ink is pronounced "re-dink" which is pretty much impossible to convince to an English-only speaker. Hence the confusion between "read-i-ly" or "rea-di-ly" or "rea-dil-y". I'm no authority but I believe the middle one would be British English, well-enunciated, and the last American, employing a dark L. But I wouldn't be surprised if most dictionaries listed the first. DAVilla 23:48, 24 November 2006 (UTC)
So if I understand correctly, you are saying: you can add both, but it is difficult. I do not really agree French has no stress. Less explicit perhaps, but it is there. For example, in avancer, obviously the last syllable is stressed, whereas in avance, it is the second. This is the last in both cases, but I am unsure whether this is a rule. henne 17:55, 25 November 2006 (UTC)
Stress in French is not phonemic, and in practice it is almost even anyway. avancer is certainly not stressed on the final syllable. French often sounds like it has final-syllable stress to English-speakers (I don't know whether you are one or not) because its equal stress pattern sounds strange when you're used to the penultimate stress of most English words. That's why when French words are adopted into English they often take final stress, especially in America (e.g. cliché, passé etc.). Widsith 14:48, 27 November 2006 (UTC)
This is new for me. You could have looked on my user page to see that I am Dutch speaking, French being, along with English, secondary languages. Ah, I wished everybody used the {{Babel}} template on their user page. Connel, maybe this is something to include in your standard welcome talk?
Anyway, Ok, so French has no stress you claim. Then indeed, marking syllables makes sense, but is it useful? Should syllables always be marked? I often add pronunciation for Dutch entries, which has primary stress, but no secondary. Should I add it there? Should I add syllable markers for non-stressed syllables? henne 12:33, 30 November 2006 (UTC)
Although I use {{welcome}}, {{pediawelcome}} and {{welcomeip}}, I try to let other people play around with the wording. Whenever I mess with them I get too many complaints. Please be bold! --Connel MacKenzie 23:40, 30 November 2006 (UTC)
I had thought French does have stress (in comparison to Japanese for instance) but that it wasn't part of the pronunciation of the word since the stress pattern can change without changing the meaning of the word. As you say, it's not phonemic. I remember a French friend saying some simple sentence that none of us got because of this, and we laughed at ourselves when we figured out why.
That doesn't mean avancer and avance, when said alone, aren't stressed at the end. Of course, not speaking French, I really wouldn't know. I've never even been able to figure out if the final R is pronounced or not. DAVilla 14:11, 30 November 2006 (UTC)
Some linguists treat the . as its own phoneme, since it can affect the meaning of a word in some cases. This makes it important. However, for most non-linguists it's difficult to properly locate that phoneme correctly in spoken English, and that's why I generally don't mark them when I add IPA to English entries. For Latin, however, I routinely include them. For English, stress is much easier for a native speaker to hear, so I add those in English (as well as Latin), though there are cases where it's difficult to decide whetehr there is secondary stress or not, particularly in compound words were there is even stress on two different syllables. --EncycloPetey 00:09, 28 November 2006 (UTC)
This would lead me to not include them, unless necessary, e.g. in Czech syllable-forming r or l (I can’t think of a word now)? henne 12:33, 30 November 2006 (UTC)

UK pronunciations[edit]

I am increasingly uncomfortable with the use of RP as a "standard" UK accent. The problem is that RP is now so old-fashioned that we are effectively giving people misleading information. For example, the vowel in man is routinely marked as /æ/ even though almost everyone in Britain including the Queen and BBC newsreaders, now uses /a/, and /æ/ sounds like some kind of incredibly old-fashioned 50s pronunciation. I personally think Wiktionary should mark (UK) rather than (RP), and take as the standard something a bit more like w:Estuary English. Specifically I suggest:

  1. /æ/ changes to /a/ (I notice the OED have already made this change)
  2. /r/ changes to /ɹ/ (this has de facto already happened with a lot of editors)
  3. /aɪ/ changes to /ʌɪ/ (again this is in keeping with the OED)

Those are the three most obvious ones that spring to mind. I am aware that this may be constituted as original research but I think the current situation is unhelpful to our users. Widsith 10:16, 8 January 2008 (UTC)

I agree with you, RP is never used anymore, except for the humour :). There are, I assume - knowing nothing about how pronunciation is transcribed, accents within UK English that should have different pronunciation sections for some words, maybe it would be acceptable to have a UK label for the cases where the pronounciation is similar from everyone, and more specific labels (including RP) for other pronunciation if there is a need for it. I don't think that Estuary English has any more claim to be `the accent` than any of the other regions - there are to my ear accents that sound more neutral. Conrad.Irwin 13:16, 8 January 2008 (UTC)
Yeah, obviously there are lots of accents in any country, but there is still a "standard" which is useful for foreign learners. In the UK that's a kind of London/south-east accent, which is what I want to represent. But if you can think of more neutral accents then please suggest some! Widsith 13:48, 8 January 2008 (UTC)
Let's put as many regional accents in the pronunciation section as possible. There's no use in removing RP pronunciations just because they're RP. However, tagging something with "UK" should link to the neutral pronunciation, maybe in an appendix or to the Wikipedia page w:Estuary English. Slightly off topic, maybe we could experiment with one page (why not hinder?) and add as many regional pronunciations as possible to that page. I don't think there's any accent more neutral than Estuary English. --Keene 13:55, 8 January 2008 (UTC)
Yes, I agree that we can and should accommodate lots of regional accents. But my concern is over how we define the most neutral "standard" for the UK. I agree with you that it's probably a kind of Estuary English. Widsith 14:04, 8 January 2008 (UTC)
Certainly a step in the right direction. How now brown cow an'all that has pretty much disappeared, innit. I've always been accutely aware that more than 50% of the UK population speak with one of the many "northern" accents. Not to mention, the alarming number of L2 learners who visit UK for the first time, and think they've landed in the wrong country !! - Algrif 14:17, 8 January 2008 (UTC)
I agree with Keene (13:55, 8 January 2008 (UTC)) to the extent that we should have as many accents as possible, including RP. And, although I don't know Ukogbanian accents, I agree that if RP is not a common UK accent, then a RP pronunciation should be supplemented with a more common UK accent, which should be UK, with the RP accent marked RP, not UK. But I think that if the only UK pronunciation we have for a word is RP, then it can be marked UK (or RP), inasmuch as it is a UK accent.—msh210 17:39, 8 January 2008 (UTC)
Good idea, I like that solution. Widsith 08:52, 9 January 2008 (UTC)
A few observations: Firstly, I (not to suggest that I am alone) still use /æ/ - although in that specific word, it's /æː/, and there is a significant enough number of people and accents with the bad-lad split in the UK for it to be relevant, I think. And this leads me to my second observation: Which Estuary English? I speak with an Estuary English accent, but there is so much variation and, as yet, little standardisation - especially as far as phonemic transcription is concerned - that I'm rather uncomfortable with it being our default. RP is outmoded, sure, but the proposed changes are not significant enough (and nobody for a minute, I hope, proposes transcribing glottal stops or vocalised Ls as phonemes) for it to merit dropping RP as our standard. I can be convinced elsewise on that, mind. That said, I'd really like Estuary English to be included in our pronunciation sections (not to mention numerous other regional accents, like the rest of you), especially with audio, which is where it really makes the difference, I think.
Miscellaneous other: I don't use [u] except before l, the elsewhere-phone is instead more like [ʉ], and this is again common in EE - do we want to transcribe /ʉ/ for EE? And as for /ɹ/.. May we please, please use this for all (relevant) English dialects? It's fine for a monolingual dictionary to use /r/, but in our multilingual one, it doesn't make sense to conflate it unnecessarily with Spanish <rr> and the like. --Wytukaze 20:07, 15 January 2008 (UTC)
Do you really use [æ]? I'd be amazed, but anyway the solution above means that we will not lose RP so it will still be there - I certainly don't think it can be called a common phone anymore, although I noticed the Queen still uses it in her speech. The bad-lad split is a separate issue – that is to do with short versus long vowels (received pronunciation /bæːd/ and /læd/ compared to modern [baːd], [lad] invalid IPA characters (][)). I don't think anyone has suggested using [u] so I'm not quite sure what you're saying there. Widsith 09:36, 16 January 2008 (UTC)
Yeah, I really do. Common for EE from where I grew up (north of London, Bedford-Milton Keynes area) although the short [æ] is also sometimes [a] - the long [æː] doesn't appear as [aː]. (Incidentally, I don't think RP traditionally underwent the split.) In any case, I'm fine with using /a/ (and /aː/ too) for the EE phoneme, I just wanted to note that it's not dead yet. As for [u], has anyone suggested not using it? That's what I'm getting at. It's not my default phone, and I'm wondering if that's common enough. A lot of EE speakers do have a central equivalent of RP [u], as do a lot of Australians, so, as is the norm, shouldn't we express the phoneme using the elsewhere-phone?
My reservations to do with the lack of a transcription standard for EE remain. Do we transcribe <think> as /fɪŋk/ (I don't use this, but it's popular among modern EE speakers)? Do we note yod-coalescence phonemically (/ʧʉːzdɛɪ/)? It's not entirely universal among EE forms, but I'd say it's pretty much a given. And diphthongs! We have issues here. /aʊ/ is most commonly [æʊ], I've seen [ɑɪ] preferred to [ʌɪ] because the latter is equivalent to RP [eɪ] for a huge amount of speakers (and my /ɛɪ/, another EE pronunciation variant, in the 'Tuesday' above). And L-dropping, while not necessarily phonemic itself, causes some interesting mergers and splits for many of us, some of them probably minimally phonemic - <wholly> [hɒʊli] or similar (and <wholly> might get a geminate L ([hɒʊo.li]) here, distinguishing it with <holey> as well, but this is unlikely except in very careful speech) vs holy [hʌʊli], and <bald> and <bold> both as something like [bɒʊod] (/bəʊld/? /bʌʊld/?). More London-based speakers might make <fail> [fæo] and <mile> [mɑo], merging them with the reflex of /æl/ and (perhaps) /ɑːl/ - <canal> [kəˈnæo] and <marl> [mɑ(ː)o]. There's a pretty definite split of <coolly> and <truly> for speakers all over the place - [kuoli] vs [t(ʃ)ɹʉːli]. Then there's RP /ʊə/, which is hardly ever not merged with /ɔː/ nowadays - except in certain positions and certain words; <jury> is not often [ʤɔːɹi]. I'm not especially attached to keeping RP - other than for the fact it is still taught and used - but prefer it to moving to EE as our default British dialect. If I'm wrong, and there is in fact a standard - a commonly accepted and used standard - for transcribing EE phonemically, do tell, I'd be very interested. --Wytukaze 19:45, 22 January 2008 (UTC)
I don't know the answers to these questions! You're right of course that there's a whole load more potential changes we could make, but that's why I only picked the three mentioned above, which I thought were the most pressing and least controversial. Widsith 09:38, 23 January 2008 (UTC)
Well, see, that's my point; we need to sort out this sorta thing out before we employ it as our UK standard. And, to be frank (not that I haven't been being frank before - sorry about that), those changes are probably the least pressing - they mean transcribing everything exactly as RP, but using different symbols. Laudable, perhaps, for their increased accuracy, but surely where it counts is where the two forms diverge markedly - splits (like /ɒʊ/ from /əʊ/), mergers (like /ʊə/ with /ɔː/, arguably /θ/ with /f/ (and /ð/ with /d/?), arguably /tj/ with /ʧ/ and /dj/ with /ʤ/) and different forms (like /ɪn/ for /ɪŋ/, that is, <-ing> in <run-n-ing>, but, again, this isn't universal, but it's probably got the upper hand at the mo). Some input from others who might know EE (or, hell, anyone with an opinion - this is a wiki, after all) would be great, actually; I don't want it to be as if I'm berating you, Widsith, that ain't my aim. You're right enough that Estuary needs a presence here (and it needs to be adopted, eventually, by British dictionaries for pronunciation keys), I'm just uncomfortable with us striking out in the dark here. I'd rather transcribe UK (RP) and UK (EE) than ever have UK alone (and RP as a transcription standard is gonna have more people, more contributors, familiar with it). But we need to make ourselves a guide, preferably based on published - and freely available - research, rather than just slapping in what are basically the same pronunciations with a new name and a lack of clarification. So yeah, someone else weigh in. --Wytukaze 00:46, 24 January 2008 (UTC)
Yep. Just on the point of published research and otehr dictionaries, the OED as of last year has adopted two of the changes as noted above, and they also regularly transcribe both /tj/ and /tʃ/ where both are still common. Widsith 09:02, 24 January 2008 (UTC)
It's not especially obvious that they're transcribing for EE, though. --Wytukaze 00:42, 25 January 2008 (UTC)
Well, no - we're calling it that but the point is just that a mild form of EE is de facto becoming the new UK standard. Widsith 09:42, 25 January 2008 (UTC)
I weigh in with this (which somebody may have already asked): Why Estuary English? That is totally unrepresentitive of the middle-upper class, East Anglia, the Midlands, the North and the Far North as well as Wales, Scotland and N.I. It should not have supremacy over other dialects of English just because it is around the capital. I doubt very much that is fairly represents the speech the majority of British people. Harris Morgan 01:06, 25 January 2008 (UTC).
Basically just as Conrad Irwin said in the first response, and I'd like to disagree that foreigners are taught EE nowadays. RP is ALWAYS stated as the common standard for English teaching abroad and I give you links: [1][2][3]. Harris Morgan 01:16, 25 January 2008 (UTC).
That's exactly the problem. As your first link points out, "Today 'marked RP' is spoken only by members of the royal family and others from the upper classes. It is considered over-the-top by most people...". A mild form of Estuary English is de facto becoming the new UK standard as these sites themselves make clear. Widsith 09:40, 25 January 2008 (UTC)
I suppose. But Estuary English definitely reflects Southern accents (miwk, bāth etc.) that is why I think it is unsuitable to represent the whole of the UK. Harris Morgan 12:54, 25 January 2008 (UTC).

← A good point. I mean, yeah, EE is a southern dialect (as was/is RP before it), but as the quickly emerging new standard, it is having a significant influence on dialects across the UK, especially in metropolitan areas (again, as with RP before it). Ever heard of 'Jockney'? It's the term certain media outlets like to use to describe the t-dropping and th-fronting and various other EE/Cockney developments appearing in Scottish English speech. T-dropping (or glottalisation, if you prefer) is very common in pretty much every city and large town I can think of across England. These are all EE's doing. However! (There's always a however.) You're right. It doesn't represent the whole UK. Neither does RP. GenAm doesn't represent the whole US (although it comes a damn sight closer than anything we've got, I'll warrant). So, we shouldn't be giving a transcription with just the country name, we should include the dialect as well. I'd recommend UK-RP, UK-EE, US-GA, etc. Incidentally, anyone mind if I give Wiktionary:English pronunciation key a revamp? Just a bit of a reformat and cleanup, for now, in preparation for adding EE and more descriptive sections, and /ɹ/ should this big BP debate get resolved that way. I'd also like to move it to Wiktionary:English pronunciation, but I'll talk about all those changes on the BP in a little bit. And Widsith, I've had a think about this; is it okay if I draw up a quick proposal (perhaps on a subpage of my userpage) for transcribing EE and then work with you and everyone else from there? --Wytukaze 14:51, 25 January 2008 (UTC)

Sounds good, the whole reason I brought this up was so we could revamp our Pronunciation guidelines. By all means make a start! Widsith 14:55, 25 January 2008 (UTC)
Good idea. Harris Morgan 15:01, 25 January 2008 (UTC).
Me has been wearing a technical thinking cap for a bit... How easy/accurate would it be to automatically convert between different dialects, are the differences between dialects always the same for the same sounds? would it be possible using some javascript to convert between dialects, or would that present too many problems? Conrad.Irwin 17:21, 26 January 2008 (UTC)
I can't tell you if it could work technically - but assuming that all dialects come from several vowel changes (and allowing room for the irregular ones with a "template pipe trick" ({{Xɵʊ|Xɵɑ}}) I suppose it would be linguistically (as long as there is that variable for the irregular forms). The template that handles Ancient Greek, Some other Greek and Byzantine Greek already does something similar. Harris Morgan 17:45, 26 January 2008 (UTC).
Please note that for accents between which pronunciation differences can be automated, the different pronunciations are probably phonetic only, not phonemic. So, for any such set of accents, there would be just one phonemic transcription (given between slashes) and multiple phonetic transcriptions (given between square brackets). Rod (A. Smith) 21:10, 26 January 2008 (UTC)
Aye, we run into a lot of trouble with phonemic splits and mergers. No single dialect can be used to convert from, since every dialect will have a distinction another (or all of the others) do not have, and enPR is unsuitable for this purpose. The token system used for the Greek template would work for a lot of circumstances, with a lot of effort put into to disambiguating absolutely everything, but it can't work for situations where the pronunciation does not differ in a systematic way or where there are pronunciation variants within a dialect. Our solder, while horribly formatted and unclear, is a good example of this - most Brits would pronounce an /l/ but most Americans wouldn't, and then we have the choice between /ɒ/ and /əʊ/ within RP (or Estuary, come to mention it). Of course, we could just give the token-template a bunch of pronunciations and specify which dialects we want it to convert to... but then I think it'd be much more economical to just keep the system we've got (and let contributors use the lookup tables if they want to enter pronunciations outside their own dialect; I haven't forgotten about the tables, by the way (I've been busy)). --Wytukaze 19:50, 29 January 2008 (UTC)

← I had a question about correlating wiktionary w wikipedia. Currently English IPA links direct the reader to w:IPA chart for English, so any changes in how pronunciations are transcribed here, such as <æ> to <a>, need to be coordinated with that page. However, that link is used even for "phonemic" (meaning inter-dialectal) transcriptions, which cannot be deduced from it, since in such cases there is no defined dialect to follow. There's another English IPA chart on 'pedia, w:Help:Pronunciation, which is designed to be dialect neutral; it's basically an IPA version of wiktionary:English Phonemic Representation and SAMPA, which are also not dialect specific in the way they're being used here. Is a redirect in order, depending on the style of transcription? If so, does w:Help:Pronunciation need to be modified for 'pedia and wiktionary to find common ground? Kwamikagami 01:51, 3 April 2008 (UTC)

We don't need to correlate. If we find that the Wikipedia page is not in line with our conventions, we can write our own.—msh210 15:59, 3 April 2008 (UTC)

← We are on potentially dangerous ground here. I'm not sure whether this has been mentioned yet, but /a/ is the sound of Spanish/Italian "a" (as in "casa" in both those languages), and is very different from /æ/, which is the sound in English "cat" (as spoken in the south of England). It is false to say that the Queen and the BBC use /a/. This sound is used instead of /æ/ in some northern English accents and Scottish English. The Queen's (outdated) pronunciation of the vowel in "man" was closer to /e/. Changing /æ/ to /a/ would therefore be a big mistake, unless we want to adopt northern/Scottish English as our standard for UK English, because the pronunciation would end up being wrong. My understanding, although I could be wrong, is that the OED has changed /æ/ to /a/ because it no longer uses /a/ in /aɪ/ (the RP pronunciation of the word "eye") but now renders this diphthong as /ʌɪ/ (the Esturary English pronunciation of "eye"), so /a/ is now free to stand for whatever sound they say it does. However, we reserve (or rather, IPA does) /a/ for Spanish "a", so if we change /æ/ to /a/, that will make "cat" sound more like "cut" or "cart" (without the "r" pronounced), which will be incorrect. For the same reason, we need to distinguish between /r/ and /ɹ/ because the former is a Scottish/Spanish "r" (a trill or flap).

The bad-lad split is for /æ/ and /æː/ (a "short" sound versus a lengthened one), not for /æ/ and /aː/. It is common in southern British English but not heard in northern British English, Scottish English or American English (in most accents, at a guess).

There is no such thing as dialect/accent neutrality, by the way — everyone has an accent. RP, Estuary English, BBC English, etc, are all non-neutral. We should aim provide a standard (RP is the one usually adopted) but cannot provide a neutral accent.

Furthermore, most print dictionaries use RP, so this makes transcribing pronunciations into Wiktionary simpler.

Another point: we already have some concessions to EE over RP: final unstressed -y (as in "funny" and "slowly") is pronounced /ɪ/ in RP — again, imagine the Queen saying these words with a clipped "i" sound at the end — but sounds very quaint to modern ears. Almost all English speakers and US English use /i/ here (a shortened form of the "ee" sound in "me"); similarly, final unstressed /iː/ (as in "coffee") is usually pronounced shorter than this (as /i/ again) in contemporary English. Our IPA and SAMPA pronunciations take these changes into consideration.

Keene says "Let's put as many regional accents in the pronunciation section as possible." This has been proposed before in the past, but is a bad idea for space considerations (even though Wiktionary is not paper). The typical user is not interested in knowing how a word is pronounced in hundreds of regions of the UK, US, Australia, Canada, etc. They just want to know how the word is pronounced. We already give (or intend to give) standard pronunciations for each country. Regional variations can almost always be given in the form of a table on a separate page designed to explain how to "convert" one regional accent into another, and I have proposed this in the past. I don't know whether one has ever been set up. — Paul G 15:19, 30 September 2008 (UTC)

  • I disagree with your analysis. Yes, of course /a/ and /æ/ are different sounds. But what the OED is asserting is that this difference is precisely what makes UK-English man sound different from AmE man. (They transcribe the two as /man/ and /mæn/ respectively.) [æ] sounds bizarre to my ears (I'm from the South-East - prime "neutral accent" territory you might think). To me, the a-sound in Britain sounds far closer to the a-sound in Continental languages than it does to American English. Ƿidsiþ 16:40, 30 September 2008 (UTC)
Fair enough, but then how are we to explain the pronunciation guide used by the online Oxford Dictionary of English, which uses /a/ for three distinct sounds, namely the "short" a in "man" (IPA /æ/), the first vowel of the diphthong /aʊ/ (as per standard IPA) and the "long" a in "lah" (they have /aː/; standard IPA is /ɑː/). This looks to me like a simplification so that only one symbol is needed rather than three, as these three sounds are different in most dialects of English.
One way to clear this up would be to write to the Oxford University Press to enquire about the changes to the IPA transcription they use. (the online ODE's "Contact Us" page doesn't have an email address for queries about the dictionary.) My feeling is that the merging of the "a" sounds is for phonemic or typographical reasons, while the change from /aɪ/ to /ʌɪ/ reflects an actual change in pronunciation.
In any case, isn't this discussion academic, as the pronunciations we give are phonemic, not phonetic? The symbol "æ" can be made to stand for whatever sound we like; the vowel in "man" in UK and US English is different, but we can safely use "æ" to stand for both because there are no minimal pairs that would cause one or other these to be mispronounced. — Paul G 09:06, 1 October 2008 (UTC)
I think that we are in danger of over emphasising Estuary English in this discussion, regardless of how common it is (and outside the south east of England it isn't at all) it is still perceived as non-standard by the majority of English speakers. My speech is primarily a mixture of northern England and the Westcountry, where such features of EE as th fronting (/ɵ/ to /f/) are completely unrepresentative.
In terms of the three changes proposed /æ/ to /a/, /aɪ/ to /ʌɪ/, and /r/ to /ɹ/, the final one is already happening and (other than in rhymes templates, which are a separate issue) I and others are changing /r/ to /ɹ/).
/æ/ to /a/ is representative of modern British English speech than /æ/ is, and I support the change, however I have not been changing it (and indeed changing /a/ back to /æ/ as there is not consensus for it yet and /a/ does not appear in our or wikipedia's pronunciation charts as a phone used in English - this is an absolute prerequisite before using it in transcriptions.
/aɪ/ to /ʌɪ/ I don't agree with, as it's not something I hear in the majority of English speech either in London or the Westcountry. Perhaps it is a feature of EE, but as I said above I oppose making that the standard. Thryduulf 12:16, 2 October 2008 (UTC)
  • OK. I wish I'd never mentioned Estuary English...it was misleading. I never wanted EE to become standard, I only meant that the "standard" (whatever that is) is already using some EE features. It is also using some northern features -- /a/ for /ɑ:/ is much more common than it used to be. Anyway, that's beside the point. My only concern was that we should find some more modern alternative to RP, which is not a very useful guide to how English people speak, in my opinion. To answer Paul's point – I suppose you could say that this notation is more phonetic, but as I said somewhere else, transcription is not a binary system, it's a sliding scale. There are different levels of "narrowness". It's true that there are no minimal pairs between /a/ and /æ/, but that would only be an issue (to my mind) if we used both symbols within a given country's phonemic range. But we don't -- we would only use /a/ for UK and /æ/ for US. Compare /ɒ/ and /ɑ/. There are no minimal pairs between them either, but we still use both because it's helpful for users to see this basic feature of transatlantic difference. Ƿidsiþ 15:36, 2 October 2008 (UTC)

Homophones section[edit]

See also the archived discussion Wiktionary:Beer_parlour_archive/2007/April#Homophones_as_a_L4_header

I have started a vote (Wiktionary:Votes/2008-01/Homophones section) to establish Homophones as a distinct L4 subheader and section under ===Pronunciation===. Please note that this disucssion is independent of that vote. The vote is simply on whether or not to have the new L4 header; this discussion concerns possible layout should the proposal be approved. We would still need a final wording, and an additional vote on that wording, before any recommended format is added to WT:ELE.

Based on previous discussion (linked above), I propose the following format, should the new subheader be approved:

  • For simple cases, where homophones are pronounced the same regardless of dialect, a single-line bulleted item with the homophones in alphabetical order will suffice.
For rite, the entry and homophones have no significant variation in how they are pronounced:
* [[right]], [[wright]], [[write]]
  • For complex cases, where homophones are dependent upon dialect, region, or the particular pronunication of the entry word, a bulleted list will identify the entry pronunciation, then the homphone entries, each with a gloss as needed.
For sere, there are several pronunciations, and the homophones are dependent upon pronunciation the entry and region of the potential homophone.
*{{IPAchar|/ˈsɪə/}}: [[sear]] (UK), [[seer]] (UK)
*{{IPAchar|/siːr/}}: [[sear]] (US)
*{{IPAchar|/ˈsiːɚ/}}: [[sear]], [[seer]] (US)

Thoughts? --EncycloPetey 19:49, 27 January 2008 (UTC)

    • At the moment, this is covered by saying whether the homophone is for rhotic or non-rhotic accents; I think saying as much covers all of the above and makes it unnecessary. By the way, "seer" is not necessarily a homophone for "sear" in UK English - it can be pronounced /siːə/. — Paul G 15:31, 30 September 2008 (UTC)


I've added a hyphenation section, following what I understand to be common usage, and discussion at Talk:ELE; how does it look?

Nbarth (email) (talk) 17:45, 17 February 2008 (UTC)

Thanks for taking the initiative to get this started! I have edited the content a bit for clarity and have included a specific example to show how syllabification differs from hyphenation. I've also taken the liberty of moving the Hyphenation explanation to the end of the "other sections". Since this is the one item included in the Pronunciation section that is not actually about pronunciation, I feel it ought to be listed last in the section. --EncycloPetey 18:31, 17 February 2008 (UTC)

Pronunciation order[edit]

Pronunciations appear to be given in the order enPR, IPA, SAMPA (from looking around). I've written this into the draft policy; it makes sense to me, via the logic:

  • enPR is most intuitive and least scary, so it goes first
  • IPA next, as it is important
  • SAMPA last, and immediately following IPA, as it is auxiliary to IPA

Is this general consensus?

Nbarth (email) (talk) 22:39, 7 April 2008 (UTC)

Well, yes, but part of the reason for this is simply that the order is alphabetical. --EncycloPetey 23:07, 7 April 2008 (UTC)

Removal of ad-hoc pronunciations[edit]

I've argued about this before many years ago but now that I'm trying to learn another language I'm just going to register my complaint again: I'm fed up of having to constantly refer to IPA tables and charts to find out how to pronounce a word. Sure, SAMPA might be more ambiguous, but I'd rather take ambiguity than the ridiculous amount of time and effort it takes otherwise. I wouldn't mind if you allowed SAMPA along with IPA but the fact that you're actively removing SAMPA just boggles the mind. It just seems to completely defy the site's accessibility. The bottomline is: it's finally got on my nerves too much, so I'm going to use the billion other internet dictionaries that do have SAMPA. I just can't help but notice that wiktionary is losing out to the competition here on this particular aspect. Something i thought it should never do. 08:24, 12 April 2008 (UTC)

Huh? We do allow SAMPA along with IPA; they frequently occur side-by-side in our Pronunciation sections. SAMPA is not more ambiguous, and never will be, because it in fact shows the same information as IPA does. SAMPA, done properly, is simply the IPA written with ASCII characters. The difference is only is which symbol gets used for each sound.
Could you please name some of the internet disctionaries that have SAMPA; I've not seen it in use on any major dictionary. And by the way, you might want to look up ad hoc because you don't seem to know what it means. SAMPA is not an "ad hoc" pronunciation scheme. --EncycloPetey 14:15, 12 April 2008 (UTC)
I think it more likely that the anonymous editor doesn't know quite what SAMPA is. However, it's worth noting that SAMPA is arguably more ambiguous than IPA, since each dialect of SAMPA applies only to a specific language and makes only phonemic distinctions, so that whereas an IPA transcription of English can distinguish [p] from [ph], a SAMPA transcription cannot. (I'm not sure if you're confusing SAMPA with X-SAMPA, or if you simply don't consider that relevant, since our IPA is generally an attempt at phonemic transcription anyway; but if the anonymous editor disagrees, then his or her comment makes a bit more sense. However, I'd agree with you that correct SAMPA is no more ad-hoc than correct IPA.) —RuakhTALK 17:54, 12 April 2008 (UTC)

w:IPA chart for English[edit]

Why is Estuary English not here? I thought in the discussion higher up on this page, it was concluded that Estuary English was to be included. Nwspel 23:27, 20 May 2008 (UTC)

The discussion further up the page didn't really come to a final conclusion. However, as this is a wiki you can add Estuary English pronunciation to that table (I'd encourage you to add it to the w:SAMPA chart for English also). Don't replace RP, but add a new column as EE doesn't represent the majority of UK accents any more than RP does. Thryduulf 08:51, 21 May 2008 (UTC)

SAMPA is obsolescent[edit]

Since SAMPA is an obsolescent pre-Unicode computer hack, and not likely to be known by anyone who does not also know the IPA, should its use no longer be advocated by the pronunciation guideline? —Kwamikagami 10:51, 24 January 2009 (UTC)

It is permitted. We actually use X-SAMPA rather than the strict older SAMPA, so it is an ASCII version of the IPA. It is still useful because a number of browsers and operating systems can't cope with displaying IPA characters. I can personally attest that the public-use computers at the UC Berkeley library cannot display some common IPA characters, and instead substitute the dreaded empty rectangle. --EncycloPetey 23:16, 24 January 2009 (UTC)
Still? Wow. That's pathetic. Guess it's best to keep it for now then. (I didn't mean that it shouldn't be permitted, just that I doubted it was still necessary. Guess I was wrong.) Kwamikagami 09:37, 25 January 2009 (UTC)
Has a reader ever tried to use X-SAMPA on Wiktionary? Michael Z. 2013-09-22 18:18 z

Error in instructions for non-rhotic dialects[edit]

The description of non-rhotic "better" as phonemically /ˈbɛtə(ɹ)/ is incorrect. It is phonemically /ˈbɛtəɹ/, the same as in rhotic dialects. The difference—dropping the ar except before a vowel—is predictable and therefore not phonemic. Phonemic differences are those which are not predictable, such as the lack of an [ɹ] in "bird". —Kwamikagami 10:51, 24 January 2009 (UTC)

Since there's been no objection, I've corrected better and added bird as an example of a word that truly is phonemically distinct. kwami 06:31, 12 July 2009 (UTC)

Position of audio files[edit]

I don't know if this has been talked about before, but I'd like to propose that audio files be put with the pronunciation transcriptions on the corresponding accent line. Not that this is against policy per se, but most entries, WT:ELE, and this page, list audio files with separate bullet points. Listing them together would provide better coherence compared to the current system that sometimes put the accent type in the display text of the audio link, but often doesn't. Sometimes an accent line is long (cf Wednesday which has, for a single accent, multiple pronunciations transcribed in 3 systems), but even in these cases I think it's more clear to list the audio files with the correct accent line. Sometimes there are two audio files for the same accent/pronunciation. In this case we could maybe modify {{audio}} to be able to show more than one file more cleanly. When there is no transcription for a given audio pronunciation, it might also be nice, for consistency, to note the accent with {{a}} rather than just in the audio file link text file. Thoughts? --Bequw¢τ 21:52, 19 February 2009 (UTC)

I forgot that I have such a short memory. --Bequw¢τ 17:20, 20 February 2009 (UTC)

silent t[edit]

often: /OF-tuhn/. Similar words with a silent -t- are "chasten," "fasten," "hasten," "listen," "soften," and "whistle" per Garner's newsletter. I think I have heard all of these except "listen" and "whistle" with audibly detectable "t". Is that my reading mind interfering with my hearing? Or is it actually pronounced that way by some, especially US? DCDuring TALK 16:10, 2 December 2009 (UTC)

I've heard often with a t, but don't recall hearing the others. Here's a thread on often which also touches on the other words, for your reading pleasure.​—msh210 18:31, 2 December 2009 (UTC)
Thanks. I noted that we don't seem to have an entry for spelling pronunciation (spelling pronunciation at OneLook Dictionary Search). DCDuring TALK 19:00, 2 December 2009 (UTC)
We do now; I just created it. —Aɴɢʀ (talk) 12:39, 19 January 2014 (UTC)

semi-formalized ad hoc system[edit]

how about something like this in addition to IPA/SAMPA:

  • DOSS-EE-AY ; 'do' as in dot, 'ss' as in hiss, 'ee' as in wee, 'ay' as in day: DOt,hiSS-wEE-dAY; DOSS-EE-AY

I wanted to know whether dossier was pronounced 'dossier' or 'dossiay'. IPA and SAMPA are presumably superior if you know them, but for the other 99.9% of the population who don't, it's pretty equivalent to saying "it's pronounced dfs98*(&fs09".

we need an additional system that's:

  • easy to learn
  • easy to use
  • easy to understand without having to learn
    • because it's for people who don't know a phonetic system; if they've got to learn something, they may as well learn IPA
  • reasonably accurate
    • being a little off-pronunciation is fine, as long as it can be used.

DOSS-EE-AY ; 'do' as in dot, 'ss' as in hiss, 'ee' as in wee, 'ay' as in day: DOt,hiSS,wEE,dAY; DOSS-EE-AY

breaking it down:

    • pretty easy to guess how this is pronounced. only a little thought is needed to avoid confusion (DOH-SEE-AY, for example, the DOH could be pronounced as in DOt or DOnut).
  • 'do' as in dot, 'ss' as in hiss, 'ee' as in wee, 'ay' as in day
    • elaboration, just to make sure they're pronouncing it correctly
    • word-choice could be improved piecemeal as we go along (wiki method)
  • DOt,hiSS-wEE-dAY
    • should be reasonably intuitive what we're doing here -- you read the capitals, the lower-case letters just form words that contain the sound
    • if you don't figure this out, you read a string that rhymes with the pronunciation
    • if you figure this out, you can get the pronunciation by just reading this bit
    • re-enforcing the pronunciation makes the system less confusing on first encounter
    • reading after the dot-hiss-wee-day should hopefully force correct pronunciation


  • contest
    • konTEST <-- verb
    • kon-test <-- noun


just have different sections for UK, USA, etc.

easy (to understand)!

what'd you guys think? any problems? --Arkelweis 03:01, 22 January 2011 (UTC)

We have such a secondary system on Wikipedia. The 'eye' vowel is a real mess; the common convention of 'y' gets into problems in words like kryps, tyts, fyndz etc., but if we use 'ye', then there are problems with words like vyess; also siecle and cycle would both be written syekl despite having different vowels. Dict.com uses 'ahy' to get around this: krahyps, tahyts, fahynds, vahyss, etc, but that really isn't much of an improvement. It's also hard to distinguish between the foot and food vowels in a way that is clip&paste friendly; the BBC uses 'uu' vs 'oo': fuut, food, which is easier than the IPA, but still a little weird.—kwami 11:41, 25 January 2011 (UTC)
I'd say CRYPS; 'cry' as in 'crying', 'p' as in 'pan', 's' as in 'is'; CRY,Pan,iS; CRYPS
for tights: TeyeTS; 'teye' as in 'time', 'ts' as in 'bits'; TIme,biTS; TeyeTS. And yes, you have to take care not to end up with TITS, as 'ti' as in 'time' would be more intuitive :D
FINDS; 'fi' as in 'fire', 'nds' as in 'ends'; FIre,eNDS; FINDS
food is easy, but foot has me a bit stumped. but, just find another common word that has the 'uh' sound in it --Arkelweis 19:27, 10 March 2011 (UTC)
The real problem, however, is that no matter which conventions we choose, there will be syllables which people will misread, because they will be homographs to common words with different pronunciations. —kwami 11:41, 25 January 2011 (UTC)
With care, it should be possible to find a relatively short and common word with universal pronunciation for each common (short) combination of phonemes? Then, we could get a computer script to auto-generate the intuative pronounciation guide from the IPA; in the dossier example, it could check the database for the first sylable dɒs, find nothing, check for dɒ and see dɒ=DOt, check s and see s=hiSS, and auto-generate it from there. --Arkelweis 19:27, 10 March 2011 (UTC)
For your very first gloss, 'o' as in dot, for example, there are already problems: 'cobalt' would be written KOH-bolt, but most people would read that as "co-bolt".—kwami 11:41, 25 January 2011 (UTC)
It was 'do' as in DOt btw. And, I might be mispronouncing cobalt (i'm english, if it helps) but: CO-BALT; 'co' as in 'co-op', 'bal' as in 'bald', 't' as in 'time'; CO-op,BALd,Time; CO-BALT. 'course, we'd have to check that co-op, bald, and time actually have universal pronunciations. --Arkelweis 19:27, 10 March 2011 (UTC)
Where in England? It isn't KOH-BALT in RP, it's KOH-BOLT, but people reading that would tend to rhyme it with bolt. That's the problem with any respelling system. kwami 09:52, 31 March 2011 (UTC)
Hampshire (or 'amshAAARRR', as we say here). you've lost me on KOH-BOLT btw... CO-BALT seems to be right, given the pronunciation clues given (wouldn't rhyme with bolt, for example: would sound similar to the beginning of the country BALTimore, and rhyme with fault)? --Arkelweis 02:37, 5 May 2011 (UTC)
But A isn't the vowel in 'Baltimore', it's the vowel in 'Alice'. So CO-balt would be completely wrong. (I can hardly even say it.) In our scheme, it's written O as in 'hot', so the proper respelling would be KOH-bolt. But because the English word 'bolt' isn't pronounced with the O vowel, but with the OH vowel, most people would misread KOH-bolt /ˈkoʊbɒlt/ as KOH-bohlt /ˈkoʊboʊlt/. That's the problem with respelling systems: there will always be interference from English words which mismatch the system, no matter what you come up with. Thus there will always be words we can't respell. So we'll end up with a system that only works some of the time. kwami 07:15, 8 May 2011 (UTC)
is this how you're pronouncing it? CO-BALT; co as in co-operate, bal as in bald, t as in time; CO-operate,BALd,Time; CO-BALT --Arkelweis 14:22, 11 May 2011 (UTC)
You and kwami are talking at cross-purposes. He is assuming that any desirable representation scheme would be consistent; for example, if the letter "a" represents the vowel in "Alice", then it can't also represent the vowel in "bald", and if "o" represents the vowel in "dot", then it must also represent the second vowel in "cobalt". In particular, I don't think he's recognized that you're proposing long explanations like "DOSS-EE-AY ; 'do' as in dot, 'ss' as in hiss, 'ee' as in wee, 'ay' as in day: DOt,hiSS-wEE-dAY; DOSS-EE-AY". That took me a while to understand; I, for one, thought you were proposing just "DOSS-EE-AY", and the rest of the explanation was just for readers at Wiktionary talk:Pronunciation. But on the other hand, you write below that "the actual implimentation could be done by a computer based on the IPA transcript", which is not true for a complex, inconsistent system like you describe, one where (as you admit above) "a little thought is needed to avoid confusion." —RuakhTALK 20:50, 11 May 2011 (UTC)
Ah, cheers if that's what's happening. Yes, the long-explanation is part of what the user'd see. I think I'll write up a clearer 'spec' in a bit. however, when I said it could be computer generated: for dossier, if someone's put the IPA in (dɒs.i.eɪ) then a script could look up 'dɒs', maybe fail to find it, then look up 'dɒ' and 's' seperately; as long as the database contains dɒ as in DOt and s as in hiSS then it'd be trivial to have it auto-generate DOSS-****; 'do' as in DOt, SS as in hiSS****** DOt,hiSS****; DOSS-****. The IPA for the 'o' in cobalt would be different, hence the easy-to-understand explanation of how to pronounce it would be different.
one problem with this is that, as said above, tights could easily end up as TITS; maybe just run through a swear-filter, and if it matches mark it down for manual creation? there should only be a few?
that just leaves us having to write the script and database, and do a few manually, for an intuitive easy-to-understand-without-learning-it system for use alongside IPA and SAMPA --Arkelweis 15:01, 20 May 2011 (UTC)
Even worse, what in the world would you do with David Bowie? The dispreferred pronunciation is BOW-ee, but many people would read that as the preferred pronunciation BOH-ee, so it would be difficult to distinguish them.
BO-EE; 'Bo' as in bow and arrow, 'ee' as in 'bee'; BOw and arrow,bEE; BO-EE --Arkelweis 19:27, 10 March 2011 (UTC)
So how would you write the 2nd pronunciation? kwami 09:52, 31 March 2011 (UTC)
how is BOW-ee pronounced? if you want to use IPA in your examples, i'll slog through till i know how to pronounce it --Arkelweis 02:37, 5 May 2011 (UTC)
BOW-ee would be /ˈbaʊ.iː/, as in take a bow. But it looks like /ˈboʊ.iː/, as in tie a bow, which would be written BOH-ee. kwami 07:15, 8 May 2011 (UTC)
"which would be written BOH-ee" under which system? if it's bow as in 'take a bow', it'd be BOW-EE; BOW as in 'take a bow', 'ee' as in 'bee'; BOW-bEE; BOW-EE --Arkelweis 14:22, 11 May 2011 (UTC)
The consequence of this is that there will always be a substantial fraction of words which cannot be adequately transcribed in a respelling system. See the quote at fauxnetic. —kwami 11:41, 25 January 2011 (UTC)
First, WT:ELE does say "it is not wrong to use an arbitrary representation if that’s all you know and there is an important point to be made. For the word reject, one could have /RE-ject/ and /re-JECT/ to make the important distinction between the pronunciations of the noun and verb forms. It may not be standard, but neither is it wrong. Whenever possible, however, such ad hoc pronunciations should be replaced with one in an unambiguous system, such as IPA". So if you can add ad-hoc pronunciation to an entry that has no pronunciation, please do! But that doesn't really address you question, which is what we should use by default, for usability. IPA has the advantages of being relatively accurate, relatively unambiguous, common, relatively easy to learn by laymen (as easy as any other system AFAICT). But for laymen who don't want to learn IPA we do have another system in place, for English entries only. We call it enPR (for "English Phonemic Representation"). See [[WT:enPR].​—msh210 (talk) 16:58, 25 January 2011 (UTC)

I don't think wikipedia's re-spelling or enPR are particularly intuitive; for a start, they have non-standard symbols, and it should be relatively clear that non-standard symbols will not have an intuitive pronunciation. If they're easier than IPA, then lets include them aswell, but we still need one intuitive, no-learning-required system alongside the IPA etc.

You know, an alternative might be simply to have a program auto-generate and include an audio clip of the pronunciation based off of any IPA/SAMPA that's present. Or both?

--Arkelweis 19:27, 10 March 2011 (UTC)

  • Rather than spending time trying to work out the details of yet another system, why don't you just spend ten minutes learning how IPA works? It's not as complicated as you seem to think it is. And it was actually designed specifically to solve the problems that you're talking about. Ƿidsiþ 06:58, 11 March 2011 (UTC)
    • because it takes more than 10 minutes to learn.
    • because even if it didn't, IPA's still no use for the majority of people, who haven't spent that 10 minutes learning it (so it doesn't do a very good job if it's trying to solve the same problems i'm talking about).
    • because 100% of the effort for this way could be done at system creation: rendering into the format could be auto-generate from the IPA, and people will be able to use it with no learning (bearing in mind that the vast majority of people haven't/won't/don't want to learn a system like IPA).
    • so: because i'm trying to fix the problem in general, not just for me --Arkelweis 19:46, 28 March 2011 (UTC)
      • Regarding computer-generating sound files from IPA: I believe there's a program that does this. But any such sound file will not completely accurately represent pronunciation unless the IPA is very narrow, which is hard to write accurately. (And not even then, I'd guess.)​—msh210 (talk) 16:04, 29 March 2011 (UTC)
You're right; the IPA does take more than 10 minutes to learn. It takes (in my experience of teaching it to people) about 45 minutes to learn (passively, i.e. to recognize the symbols used for one's own native language - learning it actively, to generate transcriptions oneself, takes a few hours). But the problem with your solution is the same as every attempt to find a "more intuitive" alternative to the IPA: when you actually get down to the details of implementing one, it's always either terribly ambiguous or at least as complicated as the IPA itself. —Angr 10:18, 25 April 2011 (UTC)
complicated is ok, as long as it doesn't require the reader to learn the system; I feel the above could achieve that. As for ambiguous, I don't see that it needs to be. Bear in mind that 'succinct' or 'efficient to use once learned' aren't objectives, giving us some leeway that other IPA replacements don't have
specifically, rather than trying to represent each sound with one glyph (requiring more than 26, i.e. some 'alien' glyphs, to be used, thus requiring the system to be learned) we'll use pre-existing glyphs or glyph clusters, with a common word used to indicate which of the many possible pronounciations is meant; this is what makes it intuative and possible to read without learning the system, whilst not being ambiguous, BUT it's also what makes it long-winded and cumbersome.
to clarify, if you're going to be involved in linguistics a lot, IPA will remain better than this system; BUT, if you just very occasionally want to look up the pronunciation, this system'd be better as it requires zero learning (BTW, I'd suggest we have IPA and SAMPA and this system: people who know IPA can use IPA, people who know SAMPA can use SAMPA, people who know both can use their favorite, and people who know neither, who're currently stuffed, can use this system.) --Arkelweis 02:37, 5 May 2011 (UTC)
The problem is, this system doesn't require "zero learning". It may seem like it at first, but once you start implementing it, you'll find it requires almost as much learning as enPR. These ad hoc systems are never really as intuitive as they seem at first blush. —Angr 07:51, 8 May 2011 (UTC)
care to give an example of what'd have to be learned in order to read this system? the actual implimentation could be done by a computer based on the IPA transcript, so no need to learn to write it, just as long as it's intuative to read --Arkelweis 14:22, 11 May 2011 (UTC)

This system seems completely useless

  • a) for people who are not already familiar with English pronunciation rules, and
  • b) for words from other languages than English

-- 21:35, 20 April 2011 (UTC)

people who aren't familiar with English pronunciation can read their own, native wiktionary (after all, this wiktionary's descriptions are in English); failing that, the minority who can read, but not pronounce, English, can use the IPA or SAMPA. /some/ Words from languages other than English can be expressed by this system, and for all others you can either give the naturalized English pronunciation or a close approximation:
  • watashi; WA-TA-SHE* ; 'wa' as in wank, 'ta' as in tampon, 'she' as in sheet* ; WAnk,TAmpon,SHEet*; WA-TA-SHE* (SHE is approximation of し, which has no exact pronunciation; see し or IPA ɕi for exact sound)
having said that, if there's no English equivalent, then I'm doubtful that English speakers will be able to learn to pronounce it from text, regardless of which system's used. --Arkelweis 02:37, 5 May 2011 (UTC)

By writing "WAnk" as example, do you mean /wæŋk/ or /weɪŋk/ (wank)? In other words, the proposed system is inaccurate even for English speakers. Unless you specify the pronounciation for almost every word you use as example (using what? IPA, SAMPA, more words in English?) 18:45, 20 September 2013 (UTC)

Yes, the proposal assumes our readers are illiterate . A more consistent respelling system can be found at WP-en.

And no, we can't use it for other languages . You said with watashi that the SH wasn't quite right , but the W, A, T, E weren't quite right either . kwami (talk) 05:10, 21 September 2013 (UTC)

I’d be in favour of using such transcriptions to make pronunciation more accessible. But let’s do it right. For goodness’ sake, let’s not create another novel, proprietary system! If there are any standardized systems, let’s choose the best one. Failing that, let’s use Wikipedia’sMichael Z. 2013-09-22 18:29 z

If anyone has doubts about the value of this: at least one study showed that both linguists and non-linguists using a non-phonemic respelling (like noc-tiv-i-gant for noctivigant) took less time and made fewer pronunciation errors than those using a phonemic respelling (nok-tiv-ə-gənt), who bettered those using IPA (/nɒkˈtɪvəgənt/).
 Michael Z. 2013-09-22 19:11 z

Yeah, wikipedia's seems ok (except for the ə )... I'm not even sure you'd neccesarily have to differentiate between ə (About) and a (trAp)? for those who like IPA, note that WP's system seems to be IPA with different (more intuative) symbols; i.e., there's a more-or-less one-to-one mapping from WP's repelling system to IPA. Tho note the comments from above (copied below)

We have such a secondary system on Wikipedia. The 'eye' vowel is a real mess; the common convention of 'y' gets into problems in words like kryps, tyts, fyndz etc., but if we use 'ye', then there are problems with words like vyess; also siecle and cycle would both be written syekl despite having different vowels. Dict.com uses 'ahy' to get around this: krahyps, tahyts, fahynds, vahyss, etc, but that really isn't much of an improvement. It's also hard to distinguish between the foot and food vowels in a way that is clip&paste friendly; the BBC uses 'uu' vs 'oo': fuut, food, which is easier than the IPA, but still a little weird.(—kwami 11:41, 25 January 2011 (UTC))

you know, using WP's system and table of examples with my approach, you'd get:

dos-ee-ay; 'do' as in odd+lot, 'ss' as in see, 'ee' as in fleece, 'ay' as in face: oDD,lOt,See -- flEEce -- fAce ; dos-ee-ay

or possibly

dos-ee-ay; 'do' as in dot, 's' as in see, 'ee' as in see, ay as in face ; dot,see--see--face ; dos-ee-ay

which seems workable:

  • ++ end result is as, or only just less, phonetic as IPA/SAMPA
  • ++ end result is more intuitivly readable: dos-ee-ay vs. /ˈdɒs.i.eɪ/ or /"dQs.I.eI/ (or dossier, for that matter)
  • ++ can be automatically translated from IPA by script
  • -- requires writing of a table of common sounds and common, universally-pronounced example-words (not just odd,lot, but d+o = dot)
  • -- contains 'ə'
  • -- kwami's comments above

--Arkelweis (talk) 03:53, 31 October 2013 (UTC)


Is there any linguistic consensus where English words are broken into spoken syllables? Here is the example inexorably /ɪˈnɛks.ər.ə.bli/. To me as a non-native English speaker, this looks very random, without any logic, but with influence of the hyphenation rule in·ex·o·ra·bly. But why /ər.ə/ and not /ə.rə/? Merriam-Webster gives /ɪˈnɛk.sə.rə.bəl/, /-ə.bli/ in probably, and /bɛ.tər/ for better, this is also what I would consider phonetically straightforward (though why ɛk.sə but not əb.li?). I think the most important rule in hyphenation is that stressed short syllables must keep one consonant at the end, but that is mostly an orthographic or morphemic rule, not a phonetic one, and is not followed at Merriam-Webster. --Androl (talk) 10:48, 14 June 2016 (UTC)

No, there really isn't. The rules for splitting a word across a line break don't have much to do with phonological syllables anyway, but even phonologically there isn't much consensus on where syllable breaks are. There is some evidence that single consonants after stressed lax vowels in English belong to both the previous syllable and the following syllable, e.g. the /p/ in happy belongs to both the first and the second syllables. This is why I'm reluctant to add syllable boundaries to phonetic transcriptions of English, because it often isn't clear where the syllable boundary is. Both /hæ.pi/ and /hæp.i/ fail to capture ambisyllabicity, and no one would understand "/hæṗi/". And /hæp.pi/ would be misinterpreted as implying that the consonant is long. As for ɛk.sə vs. ə.bli, keep in mind that /bl/ is a possible onset cluster in English while /ks/ isn't. —Aɴɢʀ (talk) 16:01, 14 June 2016 (UTC)
  • Actually, although Angr is right that syllable breaks are quite debatable in English, in practice there is a convention when it comes to transcription: the so-called Maximal Onset Principle, whereby consonants (or consonant clusters) are loaded into the onset where possible. Hence the /n/ in "inexorable" (and the /p/ in "happy") prefers to start the following consonant rather than end the previous one; similarly the /b/ and /l/ of "-ably" can happily pile into the same onset, but /k.s/ falls across the syllable boundary because /ks/ is not allowable as an initial cluster in English. I believe this principle had real phonological implications in for instance early Romance languages, but in the case of English it doesn't really make much practical difference – it's just how most linguists choose to syllabify in broad notations. Ƿidsiþ 07:55, 5 May 2017 (UTC)
    • But there is also a convention that stressed checked vowels should be followed by a consonant wherever possible, so that the /p/ of "happy" should be put in the first syllable. —Aɴɢʀ (talk) 11:05, 5 May 2017 (UTC)
  • Right. The thinking is that, because these vowels cannot appear at the end of words (barring a couple of interjections like meh), then they should not be considered to appear at the end of syllables either, though not all analysts accept this reasoning. Like I say, I don't think it really matters much, my point was only that when it comes to lexicography, most authorities seem to have gone for the maximal onset principle – that's certainly how the OED and Collins syllabify. Ƿidsiþ 12:13, 5 May 2017 (UTC)
    • I think American dictionaries are more likely to use the "maximal stressed syllable principle"; the AHD for example transcribes happy as "hăpʹē" and petal as "pĕtʹl". (I can't figure out Merriam-Webster's pattern, as they write "ˈha-pē" for happy but "ˈpet-ᵊl" for petal. Syllable breaks at the ends of lines definitely follow the "maximal stressed syllable principle", e.g. pet·al. —Aɴɢʀ (talk) 12:23, 5 May 2017 (UTC)

Pseudo-IPA symbols ᵻ and ᵿ[edit]

In November 2015, Wikipedia decided to start using the ᵻ and ᵿ symbols in phonemic transcriptions, as described in w:Help:IPA for English. These symbols are currently rejected by Template:IPA since they are nonstandard. Since Wiktionary:Pronunciation links to w:Help:IPA for English as an example of a consistent set of IPA symbols, I was wondering whether we should make these symbols eligible for use on Wiktionary or if there are other symbols that I should use in their places. Germyb (talk) 07:01, 19 April 2017 (UTC)

MediaWiki_talk:Edittools#⟨ᵻ⟩ and ⟨ᵿ⟩ might be of interest. —suzukaze (tc) 07:02, 19 April 2017 (UTC)
Thanks. That answers my question about what is currently being done. If we're going to point to w:Help:IPA for English as a reference, it seems like this difference is something we should mention in the documentation for the sake of new users. But that's just my opinion. Germyb (talk) 23:35, 19 April 2017 (UTC)
I personally think they're a very good idea, especially for UK English, but other editors have been removing them from entries. It's one of many issues that are not really resolved in our pronunciation policy. Ƿidsiþ 07:04, 19 April 2017 (UTC)
Would it be appropriate to use {{a|non-merged vowel}} and {{a|merged vowel}} similarly to how Wiktionary:Pronunciation suggests handling the rhoticity distinction? From what I can tell, that only applies to ᵻ. I don't know if there is a similar approach for ᵿ. Germyb (talk) 09:08, 19 April 2017 (UTC)
My opinion of these symbols remains the same as it was last December, as expressed in the link Suzukaze provided above. —Aɴɢʀ (talk) 12:24, 19 April 2017 (UTC)