Help talk:IPA/English

For requests for transcription, see Wikipedia:Reference desk/Language.

Inclusion of /ts/ as a marginal phoneme and removal of /ʔ/

/ʔ/ is an entirely paralinguistic sound and "uh-oh" is not a valid word to base the inclusion of a marginal phoneme around. However, seeing and /ts/ is a common marginal phoneme in words like "tsar" or "Mozart", including it would probably be valid. Plexus96 (talk) 14:36, 17 May 2024 (UTC)

/ʔ/ is included for Hawaiian loans. It's illustrated by uh-oh simply because it's one of the most common and intuitive ways to illustrate the sound; it doesn't mean it's only used in paralanguage.

/t/ and /s/ are already phonemes so there's no need to list /ts/ separately. Nardog (talk) 00:42, 18 May 2024 (UTC)

/t/ and /ʃ/ are already phonemes so there’s no need for /tʃ/ as well…? БудетЛучше (talk) 18:15, 21 June 2024 (UTC)

Because it's a phoneme in English if you ask just about any linguist. See English phonology#Obstruents for why. Nardog (talk) 18:17, 21 June 2024 (UTC)

/ts/ behaves more like a consonant cluster, rather than a phoneme. It doesn't appear word-initially, at least not regularly (see e.g. [1]), only word-internally and -finally. Compare this with German /ts/ which can easily appear in this position, as in zu /tsuː/ or ziemlich /ˈtsiːmlɪç/. Native speakers of English constantly mispronounce those as /syː ~ suː/ and /ˈsiːmlɪk, -x, -ʃ/. Sol505000 (talk) 15:39, 23 June 2024 (UTC)

@БудетЛучше 2001:44C8:4183:F40B:5071:B1FF:FE2F:A594 (talk) 17:06, 24 August 2024 (UTC)

thanks for the explanation БудетЛучше (talk) 17:33, 24 August 2024 (UTC)

"/ʔ/ is an entirely paralinguistic sound"

It's also a common allophone across most dialects of English, particularly for /t/ 167.206.19.130 (talk) 19:56, 21 June 2024 (UTC)

That's neither here nor there. We transcribe any allophone of /t/ as /t/ because this key is diaphonemic. Nardog (talk) 20:41, 21 June 2024 (UTC)

Edit warring at Richard D'Oyly Carte

There is edit warring against MOS:DIAPHONEMIC on that page. It's Melbourne all over again, with the 'local consensus' nonsense. Sol505000 (talk) 15:30, 4 July 2024 (UTC)

Nice canvassing. As you're the main edit warrior on the page, it's a bit of a stretch - and more than a little uncivil - to disregard the well-founded arguments of others as 'nonsense'. - SchroCat (talk) 15:34, 4 July 2024 (UTC)

You don't get to call me uncivil when you so blatantly disregard MOS:DIAPHONEMIC and this guide. Take a good look at the mirror, mate. You are the one ignoring a well-established consensus in the APPROPRIATE place, which is right here. Sol505000 (talk) 20:59, 4 July 2024 (UTC)

I’m not your ‘mate’, sunshine, and yes, I’ll call out incivility and idiotic edit warring when I see it. There’s a talk page for you to use if you’re able to do so without reverting to further incivility. - SchroCat (talk) 21:10, 4 July 2024 (UTC)

@Tim riley: has raised a valid point: Alas, readers familiar with the real-world IPA cannot be expected to guess that Wikipedia has its own esoteric phonetic system in which the pronunciation symbols mean "this unless you'd rather pronounce it that". Our transcriptions are still surrounded by single slashes /…/. Readers familiar with the IPA will know that this means a phonemic transcription. However, our transcriptions are not phonemic, but diaphonemic. We do not mark them as such, though. I thought last year we had reached a consensus that we should mark the diaphonemicity of our transcriptions by surrounding them with double slashes ⫽…⫽, see Wikipedia talk:Manual of Style/Pronunciation/Archive 11#Distinction between varieties of English. Are other steps required? --mach 🙈🙉🙊 11:48, 5 July 2024 (UTC)

To be fair, transcriptions meant to represent specifically GA or RP are themselves most often in effect diaphonemic - the former for assuming a cot-caught distinction(sometimes also the strut-comma and weak vowel mergers, which are mostly absent from the Inland North, Western Pennsylvania, New England, and Eastern Canada), the latter for assuming no l-vocalization, and both for assuming no /æ/ splits. Célestine-Edelweiß (talk) 01:29, 7 July 2024 (UTC)

Mach, perhaps Nardog knows. As for the edit-war situation, I've contacted a random uninvolved administrator (simply, the most recently active one), who can perhaps help the situation. Wolfdog (talk) 01:49, 7 July 2024 (UTC)

Done. I've also created {{IPA //}} as a shorthand akin to {{angbr IPA}}.

I've replaced {{IPA|/.../}} in the Key section where they were unambiguously referring to diaphonemes rather than phonemes as far as I could find, but it can be ambiguous sometimes. (E.g. should it be "/A/ is merged with /B/ in accent X", "⫽A⫽ is merged with /B/ in accent X", or "⫽A⫽ is merged with ⫽B⫽ in accent X"?)

Further fixes in articles where our diaphonemes are mentioned are welcome. Nardog (talk) 12:10, 7 July 2024 (UTC)

That’s great, thanks a lot! --mach 🙈🙉🙊 06:48, 8 July 2024 (UTC)

Undone per request. I suggest you start an RfC if you want to go through with this. Nardog (talk) 08:54, 8 July 2024 (UTC)

I have opened a request for comment to discuss the matter: Wikipedia talk:WikiProject Linguistics#RfC: Should we keep our non-standard use of single slashes to enclose diaphonemic transcriptions? Wikipedia talk:WikiProject Linguistics#RfC: Should we keep delimiting diaphonemic transcriptions with single slashes? --mach 🙈🙉🙊 21:38, 8 July 2024 (UTC)

I just suggested it at that talk page, but in case it's of more interest here: since it seems that //r// is so much more of a chronic objection than the fact that our key makes vowel distinctions that AmE doesn't, perhaps the problem is rather that "don't pronounce this letter that's clearly there" is a bigger mental block than "pronounce these two letters the same way". So perhaps it might help to change the symbols we use for the diaphonemes to //ɑː(r)//, //ɔː(r)//, etc., explicitly parenthesising the r? Double sharp (talk) 08:17, 11 July 2024 (UTC)

Using superscript ⟨ʳ⟩ has been suggested, e.g. here. That seems less intrusive than parentheses, and has precedence in CamGEL. Nardog (talk) 08:33, 11 July 2024 (UTC)

Even better, then, if there's a precedent. I'm happy to support superscripts if they get proposed. Double sharp (talk) 08:56, 11 July 2024 (UTC)

Either solution seems fair given the insistence of some to disregard well-established conventions over the mere sight of the /r/ symbol. I support that. ~ IvanScrooge98 (talk) 10:03, 11 July 2024 (UTC)

I also am content with either option. Something to consider for the implementer, though, is that we currently transcribe a word like bearing as b|ɛər|ɪ|ŋ in this system. If we made the change, we would want to avoid the output being something like bɛəʳɪŋ, since the r here in these types of cases is not optional even in non-rhotic accents but required. Wolfdog (talk) 14:12, 11 July 2024 (UTC)

That can be automated in Lua. Can be tricky, but feasible. Nardog (talk) 14:18, 11 July 2024 (UTC)

Nobody has intervened any more. Are we going to give ⟨ʳ⟩ a try or should we open a separate dedicated discussion before? ~ IvanScrooge98 (talk) 14:17, 21 August 2024 (UTC)

I would wait till the double slash RfC is closed. Nardog (talk) 02:15, 25 August 2024 (UTC)

i’m sure that replacements using a regex such as ([iɪʊuəɛɜʌɔæɑɒː])ɹ([iɪʊuəɛɜʌɔæɑɒː]) will suffice for finding R-liaison. БудетЛучше (talk) 18:11, 24 August 2024 (UTC)

Here's what really happened to initiate this discussion. There was a longstanding stable IPA on the Richard D'Oyly Carte page, and then User:Sol505000 edit warred there about it, but they did not get a consensus in their favor. So they brought the issue to this Talk page to see if they could get a different consensus here. -- Ssilvers (talk) 23:04, 11 July 2024 (UTC)

I have had disagreements with Sol505000 in the past, but we really should look at the situation and realize having a discussion about {{IPAc-en}} on the talk page of a specific article, when the template is used in countless pages, is out of place. ~ IvanScrooge98 (talk) 23:48, 11 July 2024 (UTC)

Shouldn't /mj/ be on the list?

I am nothing but a casual observer. I've never studied phonetics, language, speech, etc., so I have no idea if this is stupid to even bring up. If so, I offer my deepest apologies for wasting anyone's time. That being said, I saw entries for several other "letters followed by the (as I was taught in grade school) long 'U' sound," e.g., /dj/ dew, /lj/ lute, /nj/ newt, etc. Shouldn't there also be an entry for /mj/ as in music, mew, munificent? I have no clue if this is something valid that has been overlooked, or if I just don't know enough about my native language (all of which I'm sure could fill several VERY LARGE tomes) in this instance. Thank you for your time. Brutus6844 (talk) 18:46, 25 August 2024 (UTC)

@Brutus6844: It's because there is considerable varation in the realization of /dj/, /lj/ etc. in major varieties of English (see yod-dropping and yod-coalescence). In contrast, /mj/ stays /mj/ in all supra-regional varieties and only has yod-dropping in some traditional dialects of England (e.g. stereotypically in Norfolk), which is beyond the scope of variation that we try to capture in this pronunciation key. –Austronesier (talk) 19:56, 25 August 2024 (UTC)

/ɜːr/in British place names

I just came here as I noticed the IPA transcription for Herne Hill is given as /hɜːrn/. I checked a few other English place names and noticed the same use of /ɜːr/. The problem is that nobody says that, whether their accent is rhotic or not, in rhotic accents it is /hɜrn/ and non-rhotic it is /hɜːn/. Is there something I'm missing> Boynamedsue (talk) 07:13, 12 September 2024 (UTC)

@Boynamedsue: The article is correct. If you look at the IPA explanation at Help:IPA/English you'll see a footnote in the entry for /ɜːr/:

In Received Pronunciation, /ɜːr/ is pronounced as a lengthened schwa, [əː]. In General American, it is phonetically identical to /ər/. Some dictionaries therefore use ⟨əː, ər⟩ instead of the conventional notations ⟨ɜː, ɜr⟩. When ⟨ər⟩ is used for /ɜːr/, it is distinguished from /ər/ by marking the syllable as stressed.

Bazza 7 (talk) 08:09, 12 September 2024 (UTC)

I have seen that text, but it doesn't answer the point. In General American the pronunciation is different to /ɜːr/, and though I have never seen [əː] given for the NURSE vowel in RP (in as much as it exists), we can definitely agree it is not pronounced as /ɜːr/. As you allude to, the standard notation used for the NURSE vowel in the UK is /ɜː/. So why are we using a phonetic transcription which is not used by anybody?Boynamedsue (talk) 16:58, 12 September 2024 (UTC)

It's not a phonetic transcription, it's a diaphonemic abstraction that encodes how it is pronounced in the major accents of natively-spoken English. The choice of /ɜːr/ is a blend of the traditional way of transcribing the corresponding RP and GA sounds, viz. /ɜː/ and /ər/. We could also have chosen /☎/, but that's less intuitive than the more realistic /ɜːr/. –Austronesier (talk) 19:18, 12 September 2024 (UTC)

@Austronesier: Thanks. That looks like a more professional way of saying what I took four paragraphs to explain below. Bazza 7 (talk) 19:21, 12 September 2024 (UTC)

(edit conflict) @Boynamedsue: I think you are mistakenly assuming that /r/ always means "pronounce it as in 'rye'". When combined with other symbols, such as /ɜː/, it can be pronounced differently or not at all, as shown in the tables and notes at Help:IPA/English.

Take a look at Collins dictionary on-line, which uses /nɜːʳs/ for the pronunciation of "nurse". Its [https://blog.collinsdictionary.com/ipa-pronunciation-guide-cobuild/ guide states that /ɜːʳ/ is pronounced as in "turn" or "third", with a note that the /ʳ/ superscript shows that in Received Pronunciation the /r/ is pronounced only when it is followed by a vowel, while in General American it is always pronounced. Apart from using a superscript, this is the same as the notation Wikipedia uses, so your assertion that this is not used by anybody is not correct.

Other publications use different symbols: Cambridge and [dictionary.com https://www.dictionary.com/browse/nurse] use the notation you used above for what they call the "UK" pronunciation of "nurse" (although people in Lancashire and Norfolk may argue that is wrong). They use a different notation for the "US" pronunciation: /nɝːs/ and /nɜrs/ respectively.

Wikipedia has chose to use the symbols and symbol combinations defined Help:IPA/English. So for "Herne", /hɜːrn/ means "h" as in "hi", "ur" as in "fur" and "n" as in "nigh" (which you will see as tooltips if your device supports hovering over the IPA), regardless of which variation of English you speak. In my southern English version of British English, I do not pronounce the "r" in "fur", so Wikipedia's IPA definition for /ɜːr/ ("ur" as in "fur") is correct for me. Bazza 7 (talk) 19:18, 12 September 2024 (UTC)

Thanks to Austronesier for explaining the fact it is not actually intended to accurately reflect ANY pronunciation. I think that it probably should, but I would imagine the debate about whether to use local standard pronunciations in English-language place names was lost long ago. I am also a little worried that people who know the phonetic alphabet might take these transcriptions at face value, and it seems strange that this aspect of WP:IPA is in effect WP:OR.

Still, I doubt it's going to change, so there we have it.Boynamedsue (talk) 22:17, 12 September 2024 (UTC)

FWIW ⟨ɜːr⟩ is phonetically representative of RP when prevocalic, as in furry, occurring. It's also phonemically representative even when not prevocalic of some accents without the hurry–furry merger, e.g. Hiberno and West Country. Nardog (talk) 22:50, 12 September 2024 (UTC)

I think that probably makes it worse to a degree! We might be better off with the telephone symbol suggested by Austrioneser. If even linguistically trained people need to click on the link and read Help:IPA/English to make use of our transcription system (and there is nothing to advise anybody they meed to do this) I don't think it is benefitting anybody.Boynamedsue (talk) 06:42, 13 September 2024 (UTC)

Semi-protected edit request on 19 September 2024 — Article: Help:IPA/English

In the vowel table, switch iː and ɪ back. NoShldJ (talk) 05:46, 19 September 2024 (UTC)

 Not done: it's not clear what changes you want to be made. Please mention the specific changes in a "change X to Y" format and provide a reliable source if appropriate. Nardog (talk) 06:30, 19 September 2024 (UTC)

COMMA example?

I'm a little confused. In my ear, and I'm pretty sure in British English in most (all?) accents, also Australian and New Zealand, maybe US but I'm not sure, the A in comma is Open-mid back unrounded vowel ʌ, not schwa ə. Like the second a panda? Or koala? 125.168.22.109 (talk) 08:53, 2 October 2024 (UTC)

In my several-decades-old British English accent, comma, panda and koala all end in /ə/. A /ʌ/ would be "too long and open", if you know what I mean. Bazza 7 (talk) 09:49, 2 October 2024 (UTC)

That's what a phoneme is: an abstract category of sounds that are not in contrastive distribution with one another. Colour in isolation may be pronounced with vowels of similar qualities, but not coloured, colourful, etc., which are still obviously colour + suffix. Nardog (talk) 14:01, 2 October 2024 (UTC)

/r/-less NEAR and CURE in South African English (and RP?)

By analogy with //ɜː//, we should probably add r-less //ɪə// and //ʊə// (note that it doesn't normally vary with /ɔː/ or /ɜː/ - another reason to make it separate from //ʊər//) for Afrikaans loanwords in South African English (see South African English phonology#Vowels). I can't imagine an American calling Reeva Steenkamp /ˈstɪrnkʌmp/, or Marie Koopmans-de Wet /ˈkʊrpmʌnz də ˈvɛt/. They'd go for /i, u/ or maybe /eɪ, oʊ/ (or, unlikely, /i.ə, u.ə/, which is the best they could do to approximate the Afrikaans pronunciation). We could also think about adding /a/ and /ɔ/ to the guide (the latter is written with ⟨o⟩ in South African English phonology, but the point is that it's a shortened THOUGHT vowel, so ⟨ɔ⟩ would be perfectly fine for this guide. Also, this is how this vowel is transcribed in Afrikaans, and invariably so, to the best of my knowledge).

Articles that would (as of now) be affected by this change:

• Botha
• Chenin blanc
• Diepsloot
• Koopmans-de Wet House
• Louis Botha
• Marie Koopmans-de Wet
• P. W. Botha
• Potchefstroom
• Reeva Steenkamp
• Roodepoort
• Vereeniging
• Vredefort
• Vredenburg
• Vredendal

And probably a few more that I'm not aware of. I'm planning on adding more SAE transcriptions in the future. I think Botha is traditionally /ˈbʊətə/ also in RP.

//ɔ// could be used in Hendrik Potgieter /pɔtˈxiːtər/ (see [2]), whereas //a// could be used in the aforementioned Steenkamp /ˈstɪənkamp/ and Koopmans-de Wet /ˈkʊəpmanz də ˈvɛt/ - and also in Vredendal /ˈfrɪədəndal/ (I think Nietgedacht also has this vowel and so does Bloubosrand). Olifantsfontein has both: /ˈɔlifantsfɒnteɪn/ (see [3]). I see no reason not to add at least the first two, as SAE is spoken by about 30 million people (give or take, I don't know the exact figure). This guide transcribes phonemic distinctions that the majority of speakers of English do not have in their vowel system, like the STRUT-COMMA distinction (see [4]), or a phonemic /x/ which, as of 2024, is probably the most South African feature of this guide, as it's disappearing from Scottish English and doesn't really make a regular appearance in other dialects.

This is how I'd add them to the table.

Also, do we have the SAE labels in the IPAc-en template? Sol505000 (talk) 12:32, 17 October 2024 (UTC)

I don't see how it could be justified to add those when we no longer have a distinct FORCE. Like varieties that maintain NORTH–FORCE, SAE is not a variety large lexicographical works are available for. Whenever transcribing a word with a phonemic distinction the diaphonemic system doesn't account for, one should just not link to this key. In that vein, I'd find creating a phonetic key for SAE more sensible than adding marginal segments to this key. Nardog (talk) 04:58, 18 October 2024 (UTC)

Also, what does SAE have in idea, theatre, cruel, etc.? Nardog (talk) 05:03, 18 October 2024 (UTC)

LPD transcribes a distinct FORCE vowel as a possibility in GenAm (as /oʊr/), but I'm not sure it'd be a good idea to reintroduce it to this guide.

Yes, you mostly have to transcribe by ear, which is not a problem at least with /ɪə ʊə/. With /a/, you can go by orthography (/ʌ/ is also fine, IMO. They're close enough. Lass didn't really convince me they're not the same phoneme, but that's off-topic). For the former two, the rule is that in loanwords from Afrikaans, ⟨ee⟩ and ⟨oo⟩ (sometimes just a single ⟨e⟩ and ⟨o⟩) stand for /ɪə ʊə/. The glide is easily detectable by ear. /ɔ/ is probably the most tricky of all to transcribe. ⟨ɔː⟩ would be completely inappropriate - it makes Potgieter into Port Gieter, with wrong length. ⟨ɒ⟩ is not terribly wrong, it has the correct length, but wrong height and too little rounding. It's sometimes used instead of /ɔ/. /ʊ/ is usually too central and not rounded enough in SAE, so ⟨ʊ⟩ is as wrong as ⟨ɔː⟩. I guess that SAE /ɔ/ truly cannot be transcribed using IPAc-en.

A guide specific for SAE is doable. One for Scottish English could be created as well. If we create a guide for SAE, should that system be always used, so also in e.g. Johannesburg [dʒœʊˈhænɨsbøːɡ]?

I'm not sure. They could be smoothed in Cultivated SAE, which is basically SAE-flavored RP with little difference between the two. But I think even (some?) Cultivated speakers have at least a distinct /ɔ/, due to bilingualism. If their /ʌ/ is consistently back (and it's not in General SAE), then /a/ must also be distinct, if it's not replaced with /æ/ instead. Sol505000 (talk) 10:29, 18 October 2024 (UTC)

Starting 'a...'

I can't find the IPA for 'a' as in 'as', or 'accent', or 'advent' in the list; neither "ɑː" (cited: palm, bra, father - all longer vowel sounds), nor "ə" (cited: comma, abbot, bazaar - all more schwa sounds), fit well. What should be used, please? - MPF (talk) 00:38, 26 October 2024 (UTC)

Use /æ/. /ˈæz ˈæksɛnt ˈædvɛnt/. It's in the chart with sample words "TRAP, bag, sang, tattoo". Indefatigable (talk) 01:35, 26 October 2024 (UTC)

Indefatigable: Really? I read that as da:æ, same as de:ä, like a short -ae-, which is a very different sound. I also see /æ/ has a note "Some British sources, such as the Oxford English Dictionary, use ⟨a⟩ instead of /æ/ to transcribe this vowel. This more closely reflects the actual vowel quality in contemporary Received Pronunciation", so should it not be ⟨a⟩, which sounds far more realistic? Nobody æctuælly says an 'æ' sound when they mean æn 'a' sound? - MPF (talk) 09:43, 26 October 2024 (UTC)

I don't have formal training in linguistics, but it is my understanding that /æ/ has been the de facto standard notation for this English phoneme for over a hundred years, and it has been in use for this purpose on Wikipedia since its inception. Indefatigable (talk) 00:12, 28 October 2024 (UTC)

Indefatigable: Thanks! I'll not fight it then, even if it does look strange. - MPF (talk) 15:08, 29 October 2024 (UTC)

Indefatigable or anyone else - I've been looking (and listening to the recordings) further, and it's definitely IPA 304 Open front unrounded vowel [a] that I need, and not IPA 325 Near-open front unrounded vowel [æ], which is definitely the wrong sound. But when I put in [a], it gives me an error message? How can that be sorted so it doesn't give the error message? Thanks! - MPF (talk) 14:48, 4 November 2024 (UTC)

Have you read the prose parts of the help page? We don't represent specific sounds in transcriptions linking to this key, just abstract categories (diaphonemes). Nardog (talk) 14:59, 4 November 2024 (UTC)

Nardog - I've taken a look, and don't see anything particularly relevant there! But it is rather out of my depth, unfortunately - MPF (talk) 15:51, 4 November 2024 (UTC)