Help talk:IPA/English

Hello,

I would like to bring the issue about the current use of the IPA symbol for english. It've been demonstrated by Linguist like Geoff Lindsey (from University College London) that the current IPA symbols choosen to represent english phonems are wrong in many ways. Like for exemble the phoneme /iː/ which is not at all pronunced like a long /i/ but like /ij/ in southern brittish english. Or another examble is the vowel in "boat" depicted by the symbol /eʊ/ unlike the real used pronunciation which is /ow/

The book "English after RP" explains all of that in great details, and for a free alternative, I don't know I it's allowed to post youtube link so I don't do it but I invite anyone interested to chek the youtube channel of Geoff Lindsey, there's videos about this exact topic.

Furthermore there is the CUBE dictionnary (CUBE = current brittish english) that act as a good source using a modern proper set of IPA symbols to better discribe the way english is pronounced. Malekpe (talk) 23:10, 9 April 2025 (UTC)

English is not restricted to RP. — kwami (talk) 19:00, 10 April 2025 (UTC)

I assume that Malekpe meant the explanations of the RP correspondences of the diaphonemic signs on this page, not the actual diaphonemic system that Wikipedia uses - so what he was talking about is, indeed, restricted to RP. If he did mean the diaphonemic system, his remark was misguided not because 'English is not restricted to RP' but because that system isn't really an attempt to reflect adequately the phonetic realisation of any dialect. That said, this confusion only goes to show that the current diaphonemic system is misleading, since it looks like a phonetic transcription of an existing dialect, but is actually a set of abstract symbols that are meant to reflect all dialects simultaneously, regardless of the actual realisation. Wikipedia should adopt enPR like Wiktionary, because it does not create the misleading impression of an attempt at phonetic accuracy for those of us who do understand IPA.--Anonymous44 (talk) 23:17, 11 August 2025 (UTC)

If someone doesn't know what slashes indicate then they don't understand IPA. Nardog (talk) 23:55, 11 August 2025 (UTC)

No, slashes don't stand for diaphonemic transcription. They stand for phonemic transcription. That is not at all the same thing. Next, the very choice of using a phonemic transcription is debatable - unlike (broad) phonetic realisation, which is undeniable, it is a highly theoretical and debatable question what it is phonemic, underlying etc. The difference between phonemic form and phonetic realisation is often large and many non-native speakers would fail to figure out the difference and apply it. Finally, it is a reasonable assumption that most of our readers 'don't understand IPA', and, more importantly, that the ones who do tend to be non-native speakers who have been taught using IPA, where it was used in a broad phonetic transcription - and enIPA looks sufficiently similar to that very transcription to cause confusion. --Anonymous44 (talk) 11:55, 2 December 2025 (UTC)

Finally, it is a reasonable assumption that most of our readers 'don't understand IPA': Isn't that what this help page is for? Also, I guarantee you that practically no reader understands the CUBE system, almost certainly never having heard of it or come across it. In addition, doesn't it amount to telling everybody that doesn't speak a certain variety of England English that they're on their own? That's the problem that phonemic transcription attempts to alleviate. (I found it described in a YouTube video at https://www.youtube.com/watch?v=_Tcrv8lgLbk. The dictionary at https://cubedictionary.org/ isn't responding. Or, rather, it doesn't accept modern https: secure communication, and using old http:, it redirects to a website with an .hu domain suffix that doesn't respond.)

Maybe how CUBE would be useful here is explained in the longer video. You can certainly post a link to a YouTube video in a talk page for discussion purposes! Is it the one at https://www.youtube.com/watch?v=5OULnCCvdk8? If not, where is it? Might you please tell us where in the video the relevant portion is so we don't need to sit through the entire hour-plus of it to see what you have in mind? Largoplazo (talk) 13:43, 2 December 2025 (UTC)

No, slashes don't stand for diaphonemic transcription. They stand for phonemic transcription. – I wholeheartedly agree. We tried to amend this, but we did not reach a consensus (either way) in RfC: Should we keep delimiting diaphonemic transcriptions with single slashes? --mach 🙈🙉🙊 19:28, 2 December 2025 (UTC)

It still doesn't make our transcription of the vowels problematic unless we delve into phonemic mergers, or distinctions not covered in the guide. On the other hand, transcriptions like /nɔːrθ/ are problematic with single slashes as they suggest that the /r/ is there on the phonemic level also in non-rhotic dialects, which is not the case at all. I've heard non-rhotic Brits and Australians slip up and say things like [θɔɾt] for 'thought' when attempting a Russian or Scottish accent, countless times. The /r/ is long gone and so I'd support the switch to double slashes, also because the same applies to /hw/.

While I agree that Lindsey's system is vastly more accurate for SSB (though some changes apply cross-dialectally, e.g. FLEECE is almost universally a diphthong), it has the same problem as the reformed orthography for Australian English: ⟨ɔ⟩ is very likely to be misinterpreted as THOUGHT. The other problem is that almost nobody uses it and that probably won't change in the foreseeable future. Sol505000 (talk) 19:42, 2 December 2025 (UTC)

The fatal flaw with transcribing a specific accent is that soon there will be accusations of discrimination because we don't include my accent, and soon there will be half a dozen or more pronunciations for every entry, inconsistent and contradictory like the ridiculous system they use on Wiktionary. Because this is not a dictionary, there's a high likelihood that we'll end up banning pronunciations altogether, just as we banned Indian names and scripts (the only country so discriminated against on WP) due to the endless edit-wars that including them produced.

If you want to use double slashes or back slashes to mark our transcription as diaphonemic, I would support you, but it needs to remain diaphonemic. — kwami (talk) 20:14, 2 December 2025 (UTC)

My point is that someone who understands IPA by definition does not expect "phonetic accuracy" in transcriptions enclosed in slashes. Nardog (talk) 13:54, 9 December 2025 (UTC)

enPR is an american convention that's not readily accepted by the rest of the world. that said, i wouldn't be opposed to using it alongside the IPA instead of the current respelling system, which doesn't work for all words.

double slashes for the IPA have also been proposed, and would be more accurate, but it's been objected that anyone who needs that cue isn't likely to understand IPA anyway. i don't know how true that would turn out to be. — kwami (talk) 03:10, 12 August 2025 (UTC)

​ enPR and Respell are highly misleading to many outside of the US and most people outside of the Anglosphere. While non-native speakers should not be given priority over natives, there are many non-native users, and it is they who are more likely to need a pronunciation guide. These respellings can supplement IPA, but it should not replace it. —  AjaxSmack  15:34, 30 November 2025 (UTC)

Everyone can look up what the symbols of enPR and Respell stand for by opening this page. And such looking up is inevitable, because there is no transcription system that can be guaranteed to be understood by a lay native or non-native speaker in advance. EnPR and respell are based on the way English letters are usually pronounced, which any non-native reading English Wikipedia will already be familiar with to a great extent, so they have a good start. For non-native speakers who haven't been taught using IPA, it is, if anything, likely to be harder to understand than enPR. For those who have been taught using IPA, Wikipedia's diaphonemic system is misleading, because they are used to interpreting IPA as a broad phonetic transcription, not as a phonemic one. I.e. they may well try to pronounce the words as if this were a phonetic transcription.--Anonymous44 (talk) 11:55, 2 December 2025 (UTC)

Like for exemble the phoneme /iː/ which is not at all pronunced like a long /i/ but like /ij/ in southern brittish english. Or another examble is the vowel in "boat" depicted by the symbol /eʊ/ unlike the real used pronunciation which is /ow/. That's the whole point. We can't write a phonetic representation that covers the English spoken by more than the speakers of one variety of English. Written with correct notation, your statement would have been "Like for exemble the phoneme /iː/ which is not at all pronunced like a long [i] but like [ij] in southern brittish english. Or another examble is the vowel in "boat" depicted by the symbol /eʊ/ unlike the real used pronunciation which is [ow]." Symbols within square brackets represent the sounds that a given speaker actually produces. Symbols between slashes represent phonemes, abstract constructs put together to create a model for the actual diversity of pronunciation systems followed by speakers. So, for example, the first sound in "thin" is treated abstractly as a phoneme /θ/ that, for most English speakers, corresponds to the actual pronunciation [θ] while, for some, in some positions in a word, it's [t] or [tʰ] or [f] instead. The symbol chosen for a given phoneme is, to some extent, arbitrary. It's going to be chosen to be suggestive of the actual sound it typically corresponds to. But it isn't necessarily the actual pronunciation of any given speaker.

The alternative to this phonemic representation would be for Wikipedia to represent the actual pronunciation of a very small community of people that would then be useless to speakers from anywhere else. I think that that's what you thought was happening here, that only the speech of a small subset of the English-speaking world was being represented, but the phonemic representations cover much more of the English-speaking world. To translate that phonemic representation into the speech of a given variety of English, you just have to know how the speakers of that variety produces each of the phonemes when they speak. Largoplazo (talk) 16:23, 30 November 2025 (UTC)

The "r" sound in English is not /r/ (voiced alveolar trill), but rather /ɹ̠/, the voiced postalveolar approximant, or often for convenience generalised as /ɹ/, the voiced alveolar approximant. The voiced alveolar trill was used in older dialects of English, like high RP, but is now very far from the norm, only really appearing in Scottish dialects, and seems misplaced to list it here? I get this is explained in the footnote but it feels misinformative to generalise quite different consonants purely for the sake of convenience, as that seems to contradict the purpose of the IPA? Natejb2003 (talk) 12:26, 29 June 2025 (UTC)

Phonemic analysis isn't purely for the sake of convenience, and it doesn't contradict the purpose of the IPA. In fact it is the primary purpose of the IPA, which is clear if you read the Handbook of the IPA, pp. 27ff. Nardog (talk) 14:43, 29 June 2025 (UTC)

The linked text does not claim what you say - it only says distinguishing between phonemes is one of the purposes of the IPA, not the primary one - and in any case using broad transcription and sticking to standard letters of the Latin alphabet at the expense of phonetic accuracy is not 'phonemic analysis'. The real reason why using /r/ is acceptable is because the whole system used by Wikipedia is diaphonemic and abstract, i.e. it is not really meant to express the specific phonetic realisation of any given dialect of English, but to signal them all simultaneously. Which only goes to show, just like the previous thread, that the use of IPA for a diaphonemic transcription leads to misunderstandings. The proposals for changes are due to people not understanding that the system is diaphonemic, and the people objecting to the proposals obviously keep forgetting that fact just as the OPs do. Therefore, as I wrote above, Wikipedia should use enPR, as Wiktionary does, or something similar - a system that does not mislead readers by looking like a phonetic transcription.--Anonymous44 (talk) 23:37, 11 August 2025 (UTC)

It says From its earliest days (see appendix 4) the International Phonetic Association has aimed to provide 'a separate sign for each distinctive sound; that is, for each sound which, being used instead of another, in the same language, can change the meaning of a word'. This notion of a 'distinctive sound' is what became widely known in the twentieth century as the phoneme; /tru/ might be suitable for the English word true or the French word trou; and The term 'broad' sometimes carries the extra implication that, as far as possible, unmodified letters of the roman alphabet have been used. Nardog (talk) 23:58, 11 August 2025 (UTC)

decisions on which sounds qualify for their own letters depends on whether they're a phonemic distinction in languages, with exceptions like dentals. but it's still a phonetic alphabet; the phonemic criterion just means that few diacritics are needed for broad transcription, making for a cleaner appearance in many situations. — kwami (talk) 03:16, 12 August 2025 (UTC)

The value of each IPA letter is defined by natural classes rather than specific phonetic quality in the 1989 Principles, and that just follows in the footsteps of the first and second principles of the very first IPA of 1888. Nardog (talk) 09:45, 2 December 2025 (UTC)

Exactly! This is pointless… Compare English pronunciation of name Robert (with [ɹ̠]) to Italian, Czech, Polish, Finnish or Dutch (with [r]). The difference is significant.

What is the point of making such a mess? ~2025-39939-23 (talk) 03:59, 11 December 2025 (UTC)

Dutch /r/ can be almost anything, including the very retroflex/postalveolar approximant you can find in English. Typical Northern Standard Dutch renditions of the name 'Robert' would be [ˈɾɔʊbəɹt] (alveolar tap + pre-velar approximant), [ˈʀɔʊbəɹt] (uvular trill + PVA) and [ˈʀɔʊbəʁ̞t] (UT + uvular approximant, maybe slightly less standard) and now almost archaic (?) [ˈɾɔʊbəɾt] (with two alveolar taps). Alveolar trills are rare and their consistent usage is even rarer. Search for clips of Eddy Poelman and tell me if it sounds like normal everyday Dutch to you - to me his trills are highly unusual. Same with Polish - /r/ is far more commonly a tap/fricative/approximant in that language. In both of them (and in English), /r/ is used for convenience, though in contemporary English ⟨r⟩ has little phonetic justification (save for local dialects). Sol505000 (talk) 10:46, 15 December 2025 (UTC)

Well, you just proved my point: it is crucial to type a specific /r/ letter sound because otherwise you will never guess how to pronounce Robert in Dutch, since there are different /r/ sounds within the name. Typing ɾ or ɹ as r is just mental. It just violates IPA philosophy.

So it is important to avoid such a mess like switching English ɹ to r, otherwise you will be pronouncing incorrectly.

About Polish though you are completely wrong. Thrilled r (alveolar trill) is the only /r/ sound there. So no ɹ, r, ʀ, ʁ̞ or ʁ, just r. ~2025-41552-59 (talk) 09:04, 18 December 2025 (UTC)

The realizations I mentioned are all equally standard. Actually, you can add [ˈʁ̞ɔʊbəɹt], [ˈʁɔʊbəɹt], [ˈʁ̞ɔʊbəʁ̞t] and [ˈʁɔʊbəʁ̞t], since the target uvular trill can be undershot as a fricative/approximant (and possibly a tap, adding two further possibilities). There's no better way of covering that than using the orthographic representation ⟨r⟩. The alveolar pronunciation is very far from gone in the language and for many speakers there's a big difference in how the phoneme is realized in the syllable onset vs. the coda, similarly to German and Danish (phonetics differ, however). Many speakers dislike and disapprove of the Gooise r and/or would not accept any transcription with uvular symbols as the default ones, so that would yield another pushback. The original pronunciation is an alveolar trill/tap and the consonant still patterns as alveolar in phonology. I can't think of any better symbol that would be universally accepted by all speakers, even though I dislike it myself.

I speak Polish natively and I rarely trill it. The target pronunciation is a tap, undershot as a fricative/approximant. This statement is sourced at Polish phonology, so I don't think I'm completely wrong at all - the opposite, actually. And I've never mentioned the uvular realizations as being standard. Sol505000 (talk) 17:02, 18 December 2025 (UTC)

These are phonemic representations, not phonetic ones. As such, we could use astrological etc. symbols if we wanted. (One phonemic analysis of a Micronesian language famously did precisely that.) Typically, people use the familiar basic Latin letters where they can, and try to avoid diacritics. This makes the transcription more legible. — kwami (talk) 12:04, 15 December 2025 (UTC)

You are the one creating that nonsensical mess? So my proposition is: let’s get rid off all those “a” sounds in English like ʌ, ɑ or ə! What is the point of having all of them if we can present the lot of them as a? (with sarcastic accent)

“Typically, people use the familiar basic Latin letters where they can, and try to avoid diacritics”.

Yes! That’s why pronuncation of ie. Black Sabbath’s album Tyr is represented as /ˈtɪər/, because the inner letters are simply “basic Latin letters” (so no to both i and a because these are not basic Latin letters). ~2025-41343-05 (talk) 09:16, 18 December 2025 (UTC)

/ɑ/ vs. /ə/ is a contrast found in every single dialect that features vowel reduction, ditto for /æ/ vs. /ə/ (that, of course, still applies when /ɑ/ merges with /æ/, as it potentially does in Scotland - /ə/ remains distinct from this merged phoneme). ⟨a⟩ is typically used for either /ɑ/ or /æ/, nobody uses it for the schwa. /ʌ/ vs. /ə/ is another story, but many speakers don't have the merger. So if /ə/ is clearly a phoneme (a mid central one, not an open one), what's the problem with transcribing NEAR with ⟨ɪə⟩? After all, it's a centering diphthong, at least in older RP. The target is a schwa. Also, /ə/ doesn't necessarily symbolize a vowel at all, it can signal syllabicity of the following sonorant. And ⟨ɪ⟩ is now universally preferred to ⟨i⟩ as pre-fortis clipping makes the contrast between it and /iː/ (the famous shit vs. sheet and bitch vs. beach examples immediately come to mind here) a qualitative one in that context (when a voiceless consonant follows), with length being almost irrelevant there. Many transcriptions drop the length mark for FLEECE, which is then transcribed ⟨i⟩. No serious transcription would then conflate it with KIT. Thus, again, there's zero problem with the transcription /ɪə/ as the letters are uncontroversially used for contrastive phonemes in English (also in unstressed syllables: RP doesn't rhyme 'rabbit' with 'abbot'). And non-rhotic English usually contrasts /ɪər/ with /iːr/, so that searing is usually distinct from keyring. I think even GenAm contrasts them in this context, as KIT + /r/ vs. FLEECE + /r/. Thus, we need both in the guide. Sol505000 (talk) 17:26, 19 December 2025 (UTC)

And there's another issue: before consonants and pauses, what we transcribe /ɪər ʊər ɛər ɜːr ɔːr ɑːr aɪər aʊər ɔɪər ər/ are all pronounced without any trace of consonantal /r/ in non-rhotic dialects - that's millions (maybe even hundreds of millions) of speakers, presumably including the members of Black Sabbath itself, an English band. So that ⟨r⟩ doesn't necessarily stand for anything in that position, we just add it after every /ɪə/ to accomodate rhotic and non-rhotic speakers alike. Sol505000 (talk) 12:16, 20 December 2025 (UTC)

The strut - comma merger also happens in Rochesterian dialects as well ~2025-44015-48 (talk) 15:34, 30 December 2025 (UTC)

Yes, millions of speakers from the US, England and Wales have that merger. In fact, it is possible that the STRUT/COMMA contrast is rarer than the merger. What's almost universal is the FOOT-STRUT split, but it is often mischaracterized as a creation of a new phoneme /ʌ/, which is highly dialect-specific. Sol505000 (talk) 19:31, 30 December 2025 (UTC)

So the way I pronounce it, Ampersand and Cat's "æ"s are not the same. To me, the æ in "ampersand" sounds more like an "iæ", "ɪæ", or "ɛæ" sound to me. Try saying "ampersand" and "cat" right now. Can you hear a difference? Could you perhaps add this to the mergers or make a seperate symbol for it? ~2025-43927-71 (talk) 17:47, 30 December 2025 (UTC)

It's an environmentally conditioned allophony. For most Americans (and all Canadians AFAIK), the [ɛə ~ eə] allophone appears only before nasals, and before /r/. And this guide doesn't cover dialects in which /eə/ (the one from Middle English /a/, not the one from Middle English /aːr/, which we transcribe with ⟨ɛər⟩) is a separate phoneme. Sol505000 (talk) 18:57, 30 December 2025 (UTC)

So is it like a nasalized æ, like æ̃? ~2025-43854-80 (talk) 21:13, 30 December 2025 (UTC)

That's also a possible pronunciation (all realizations are nasalized, so [ɛ̃ə̃ ~ ẽə̃]), but long: [æ̃ː], or with slight raising [ɛ̃ː]. Non-natives mispronounce this tensed /æ/ by keeping it short, merging "man" with "men". They're always distinguished by length, AFAIK. Sol505000 (talk) 21:16, 30 December 2025 (UTC)

'Dy' exists and is standard, not just 'doo' (American)' and 'djoo' (technically, bad diction). ~2026-31781-5 (talk) 13:58, 15 January 2026 (UTC)

Nowhere do either "doo" or "djoo" appear on the page. In addition, in the IPA (which is what the page is about), "j" represents the initial sound of "you" and "yet", whereas "y" represents a vowel sound not present in most varieties of English, though you might be familiar with it if you know, for example, French ("fume") or German ("müde") or Dutch ("dus"). So the "/djuː/" that you see in the footnote does represent the pronunciation you were probably thinking of when you wrote "dy". Similarly, phonemically, "you" is represented as "/juː/" and "yet" as "/jɛt/". The letter "j" in English most commonly corresponds to the representation "/dʒ/", while "ʒ" itself corresponds to the "s" in "leisure" or "vision". Largoplazo (talk) 15:19, 15 January 2026 (UTC)