Appendix talk:English pronunciation

Archives

• Discussions from before 2017 can be found in Appendix talk:English pronunciation/Archive 1.
• Discussions from 2017 and later can be found in Appendix talk:English pronunciation/Archive 2.

CLOTH

Previously, the cloth set (moth, boss, ...) was lumped in with lot. Since the lot-cloth split has been a thing since the 17th century, I think we can safely recognize it, heh. I intend to add a cloth line soon. For example words I might suggest cloth, boss, off or office (which have a different vowel, viz. the thought vowel, from lot-vowel-having Goth, possible, profit, except for speakers who remerge lot and thought). - -sche (discuss) 21:08, 29 November 2022 (UTC)

Since cloth words are "either lot or thought, depending on the dialect", I wonder if the enPR notation should be "ŏ when like lot, ô when like thought", or what... I also wonder if enPR "ŏr" should also be split; it's a bit odd that there are two enPR notations of /ɑɹ/, two of /oɹ/, but then one that means "it's either /ɑɹ/ or /oɹ/, have fun guessing which!" - -sche (discuss) 22:32, 30 November 2022 (UTC)

I did not get around to working out how to incorporate it, but Adelpine added it to the table today (thanks!) and I've now moved it to its own line, because I think that's the clearest way to show that only some words (in the cloth cell) shift and others (in the lot cell) are not shifted. - -sche (discuss) 22:18, 28 March 2025 (UTC)

I will note that this means there now appear to be at least two places where we are prescribing one enPR symbol to represent two phonemically-contrastive sounds that we represent with different IPA symbols (and indeed have different enPR symbols for, in other environments). Namely, in addition to our longstanding ({{attn}}-flagged) use of "ŏr" to represent both /ɑɹ/ and /oɹ/, we're now prescribing "ŏ" represent both /ɑ/ and /ɔ/. - -sche (discuss) 22:18, 28 March 2025 (UTC)

The "law"/"cloth"/"thought" vowel

For completeness, should the table register /ɔ/ or /ɑ/ for American English? 161.24.0.40 11:54, 1 December 2022 (UTC)

Yeah, I think in entries we treat the General American thought and cloth vowel as being either /ɔ/ or /ɑ/ (labeled with "cot-caught merger"). Added the transcription and a note. — Eru·tuon 00:24, 2 December 2022 (UTC)

(This was added, and then I added the relevant "merger" label, bringing up the following question:)
Do we want to start listing, parenthetically and/or on a separate line within each cell (<br/>), major dialectal variants more often? For example, we could cover the northern (Boston, Inland North, ...) cot-caught merger to /ɒ/ that was discussed in the BP and which entries like thought mention, provided we're clear that all such things (e.g. cot-caught merger to either /ɑ/ or /ɒ/) should be labelled with the relevant merger and/or variety label, not presented as bare (General American). IMO it would be helpful if {{a}} allowed underscores to merely add spaces and suppress commas the way they do in {{lb}}, because it'd be helpful to be able to say, on e.g. cot, (cot–caught merger in Inland Northern American) /ɒ/, which would be clearer than (cot–caught merger, Inland Northern American) /ɒ/ which would imply /ɒ/ was used both in cot-caught merging dialects across the country and also in Inland North dialects whether they merge cot-caught or not (which is wrong). - -sche (discuss) 18:44, 5 December 2022 (UTC)

Multiple Australia and New Zealand transcriptions

This is in response to -sche's edit summary: ɔ, ɒ in the Australia and New Zealand columns are alternative transcriptions of the same phoneme. I think the ones that represent different phonemes in these columns are near with /iə/ versus /iːə/ versus /eə/, and tour with /ʉə/ versus /ʉːə/ versus /oː/. I would guess they are in somewhat free variation and used in different subvarieties because this is similar to the variation that Geoff Lindsey describes in Standard Southern British (except the eə which indicates a merger with the hair vowel not found in SSB).

Ideally we would have one transcription of each phoneme for clarity and maybe we would have some information on how to distinguish and choose between the different phonemes. It's difficult to enforce that without having some way to locate variant transcriptions. I'm considering making a program that would extract IPA transcriptions with their language and part of speech headers and their accent labels and references and put them in a database. That could be made available on a website similar to my enwikt-translations search engine. At the moment it'd be hard to consistently enforce a particular transcription system because it's so hard to locate the transcriptions that violate the rules, unless they use a very unusual symbol (like y: search on my current rudimentary IPA search engine). — Eru·tuon 01:49, 3 December 2022 (UTC)

Guess I should be clearer. /iə ʉə/ versus /iːə ʉːə/ is a smoothed versus unsmoothed vowel similar to what Lindsey describes for SSB (and there the smoothed version is probably more conservative or upper-class), whereas /iə/ versus /eə/ is whether here and hair are distinguished or merged, though the New Zealand English phonology article says the merged vowel is /iə/, not /eə/, so that needs clarifying. — Eru·tuon 01:58, 3 December 2022 (UTC)

re 'Aus/NZ borrow/forest vowels'

Discussion moved from User talk:-sche.

re: "not in free variation like e.g. Aus/NZ borrow/forest vowels seemingly are (if they're not, see their own {attn} btw!)"

The notation on Appendix:English pronunciation lists "ɔɹ, ɒɹ" because these are just different ways of writing the singular 'lot' vowel phoneme, which is inbetween [ɔ] and [ɒ] so has inconsistent representation. Maybe it would be less confusing to write it as "ɔɹ~ɒɹ" instead? – Nixinova [‌T|C] 09:41, 3 December 2022 (UTC)

So any one speaker, whichever vowel they use, will consistently pronounce all the words (lot etc, borrow/sorry etc, horror/forest etc, forum/oral etc) with that same vowel? (Or consistently pronounce all the r-coloured ones the same, at least?) Then my initial reaction is that ɔɹ~ɒɹ would indeed be clearer, or even, going a step further: should we just pick one symbol to ourselves consistently use (ɔ̞? or just pick one of ɔ/ɒ?), as Erutuon is saying above, and move mention of the imprecision/ambiguity of the vowel and the inconsistency of its representation in other sources to a footnote? For example, British ɜː may actually be əː and is represented accordingly in some other sources, but we pick one symbol to use consistently and mention the ambiguity and differing representations in a footnote. Or, as recently discussed, the General American vowel that results from the horse/hoarse merger is in between ɔɹ and oɹ, and from one source to another may be represented as either, but because those are just competing representations of the same sound rather than contrastively different possibilities the way ɑɹ and oɹ are contrastive (art vs sort) different possibilities for morrow, we picked one representation (for the post-merger sound). - -sche (discuss) 19:15, 3 December 2022 (UTC)

As a NZE speaker, I would prefer it if we solely used /ɔ/. Besides the fact that it's the symbol usually used in phonemic representations and descriptions of NZE, for me, the sound in NZE lot, cross, etc. is much closer to [ɔ] than [ɒ]; for some (broader) speakers it is probably higher than cardinal [ɒ] (note that using the RP sound as a proxy for the value of [ɒ] is misleading since it is significantly higher than cardinal [ɒ] in modern RP). Finally, the majority of our NZE pronunciations already use /ɔ/, so this would be codifying existing practice. Hazarasp (parlement · werkis) 08:11, 4 December 2022 (UTC)

/ɔ/ best represents the position of the vowel in File:New Zealand English monophthong chart.svg. All the transcriptions in w:New Zealand English phonology#Transcriptions use /ɒ/, but we're not obliged to follow them. — Eru·tuon 04:24, 5 December 2022 (UTC)

I remember coming across /ɔ/ in plenty of places (i.e. articles, books, etc.), but I may be wrong; my mind's a mess right now. Hazarasp (parlement · werkis) 12:26, 5 December 2022 (UTC)

fahrest

@Erutuon, this made me realize that although other dictionaries like MW and Collins do give forest with /ɑɹ/ as an alternate American pronunciation with no further labels, our entries frequently put it on a separate line with any of a number of inconsistent labels, ranging from "NYC" to "East Coast" to "NYC, Philadelphia, traditional Eastern New England (except for Boston), Ireland". Do you suppose we could pick a consistent but accurate label, even if it's just something prosaically descriptive like "(/ɑɹ/ instead of /oɹ/: see appendix)" or "('or' as /ɑɹ/: see appendix)" linked to a footnote/section on this page that could detail where it's used?
This suggests /ɑɹ/ is also found in words on the oral line, too, for these northish-East Coast accents. (Actually, that also fascinatingly suggests that the writer thinks there is supposed to be a distinction between oral and aural, which in practice there is because people ad-hoc shift aural because otherwise it's unusable, but which is not original / reflected in other dictionaries...) - -sche (discuss) 04:39, 5 December 2022 (UTC)

I'm not sure if all the American accents that have forest = lot (or a forest-glory distinction) can be coverable under a single transcription. There are other vowel differences in these accents that might turn up in some of the forest words and warrant multiple transcriptions of forest words. I'm basing this on the Aschmann American Dialects page. I can't think of any cases where this would affect the transcriptions at the moment though. It's pretty common for the other vowels to be schwa or other not-very-variable vowels.

The distinct forest vowel would probably have a different transcription in at least one accent, because apparently older New York City accents distinguish lot-forest (with a fronter open vowel) from father-starry (with a backer open vowel) and thought-north-force-glory, much like RP (see User:Erutuon/English low vowels), whereas newer ones merge lot-forest-father-starry. In the older accent, forest might be best transcribed /ˈfaɹəst/, contrasting with starry /ˈstɑɹi/ and glory /ˈɡlɔɹi/ (possibly with length marks or schwas in there somewhere). I can't specifically recall this, but maybe I heard it in the Marx Brothers or something from the 20th century without realizing it. (I think I often hear NYC lot with a fronter vowel than star, but that might sometimes be allophonic.)

The advantage of listing the accents in the label is that it tells where people might pronounce the word like this, which readers don't necessarily know. It's less likely to run afoul of the multiple transcription problem. In theory if someone has listed specific accents, they have checked that there are no other phonemic divergences in the word. But we should also give information about the accents that have this feature on this page.

To be more concise, we could instead use a label like "American forest-glory distinction" (or substitute different example words), or whatever term the literature uses if someone has come up with one already. — Eru·tuon 17:16, 5 December 2022 (UTC)

Monophthongization before /l/

User Whoop whoop pull up wants to monophthongize vowels before /l/ in GA. If we're going to do that, we need to reflect it here in the key. Though in this particular case, rolloff, at least according to MW online there's a clear diphthong, ROW-lof. kwami (talk) 05:40, 6 December 2022 (UTC)

(Just noting, in the interest of keeping discussion from fragmenting across multiple places, that this is discussed above in #/ol/ → Wiktionary:Beer parlour/2022/November#/ol/.) - -sche (discuss) 06:28, 6 December 2022 (UTC)

Pronunciation of "rolloff" in GA, as /ˈɹol.ɔf/; note the absence of diphthongs.

How "rolloff" would be pronounced if MW's pronunciation were still accurate for GA, as /ˈɹoʊ.lɔf/; note that this pronunciation contains a spurious diphthong that does not exist in the word as it's actually pronounced in GA.

MW is severely outdated in this instance; GA no longer pronounces "roll" as "row+l" (some regiolects still do, especially in the South, but not GA itself). Whoop whoop pull up ^{Bitching Betty ⚧️ Averted crashes} 06:47, 6 December 2022 (UTC)

"Roll" and "roll off" don't, but "rolloff" does, whereas we claimed the opposite. kwami (talk) 07:26, 6 December 2022 (UTC)

@Kwamikagami: None of those three diphthongize the "roll" vowel in GA. In certain regiolects, they do, but those aren't GA. Whoop whoop pull up ^{Bitching Betty ⚧️ Averted crashes} 07:36, 6 December 2022 (UTC)

-sche has redirected the discussion. We shouldn't be duplicating arguments. kwami (talk) 07:55, 6 December 2022 (UTC)

Add major subvarieties?

I think we should expand the table to indicate the standard (to be used in entries) realization of the phonemes in major subvarieties/dialects which I think we'd also benefit from more routinely (rather than, as at present, haphazardly) documenting in entries. One way of doing this would be to add them with labels to the table like this, though if we're adding so much text to the cells we should probably turn off the bolding. Another approach would be to explain the various subvarieties' realizations of the phonemes in footnotes like this. Putting subvarieties directly in the table would make it easier to look them over, but might make it slightly harder to look over the "main" national-standard phonemes as the table would become more cluttered. Anyone have comments / support / opposition / offers to assist? - -sche (discuss) 20:05, 10 December 2022 (UTC)

The other accents should be in separate columns, if they're in the main table, because they aren't General American. I'd like to have separate tables (like in User:Erutuon/English low vowels but with more vowels) as well because it makes it possible to put lexical sets with the same vowel phoneme close together. Probably better to put them there before putting them in the main table because people are mainly interested in the big accents and adding more columns will eventually make the table cluttered and even more difficult to edit and keep accurate (like w:International Phonetic Alphabet chart for English dialects). — Eru·tuon 01:08, 11 December 2022 (UTC)

Fair point. And I suppose it'd be easy enough to add a GenAm column (for comparison) to an American dialects table so people could still see what the corresponding GenAm phoneme(s) was/were, without this table getting as cluttered. (The space we save could go into adding columns for e.g. Irish English to this table, something else I've been thinking we should do, or maybe Indian English.) I just figure, we should have an equivalent of this table — i.e. a set of standard phonemes and footnotes about alternative representations thereof — for the dialects we often include pronunciations from, to help keep entries from using three different symbols for the same phoneme just because three different uses took the pronunciations from three different books, or from their own guesswork, that have different house notation styles (like Collins and its /bed/). - -sche (discuss) 03:09, 11 December 2022 (UTC)

Other National Varieties of English

According to the Wikipedia List of countries by English-speaking population, there is an extensive amount of countries which have more speakers of English than New Zealand (the country with the lowest population on the current chart). While I understand the current chart is made of countries where English is the most popularly spoken, shouldn't those with more speakers be represented as well? Maybe not all of them, perhaps some only a specific threshold of speakers would be less cluttered. However, I believe some of these varieties, especially Nigerian, Indian, and Pakistani English, would be very welcome to the chart, considering each has over a hundred million speakers. The aforementioned, along with the Philippines, have a larger English-speaking population than the United Kingdom, the second most populous in terms of speakers on the current chart. Surely they may deserve some form of representation? VGPaleontologist (talk) 01:55, 3 January 2023 (UTC)

As every pronunciation guide on individual pages point to this page as the "key" for IPA, this page shouldn't be used to show variations in pronunciation by region. It _should_ map IPA symbols to the sound it's supposed to represent. This is done by giving examples. The table already seems backward. Adding more columns for more countries just makes the purpose of this table less clear than it is already. It's not about population size of countries. The table should provide a standardized guide to what sound a pronunciation symbol is supposed to represent. (The variations, if they exist, of how a word is actually pronounced by region or country can be captured on the individual word pages.)

UPDATE: I should have just pointed you to the text on the page:

<quote>

For vowels in other dialects, see Wikipedia's IPA chart for English. An image of an old version of these tables is available.

For a fuller list of dialects, see [Phonetic Alphabet chart for English dialects] </unquote>'

--Liberty Miller (talk) 05:31, 8 January 2023 (UTC)

Tool

IPA Reader ( ipa-reader.xyz ). Can we use something similar for online pronunciation, using IPA?. For example {{IPA-en-reader|ˈpɪdʒən}} or {{IPA-en-reader|/ˈpɪdʒən/}} . You can see the "What Is This?" / "How Does It Work?" section in http://ipa-reader.xyz BoldLuis (talk) 21:13, 14 February 2023 (UTC)

IMO, ipa-reader.xyz is not that good, we could get actual people with those accents to say words that have those vowels or consonants. Swerup (talk) 11:54, 30 March 2023 (UTC)

Keyword and Diaphoneme

make a column for keywords that these words fall under: TRAP, BATH, PALM, LOT, CLOTH... and a column for diaphonemes??? Swerup (talk) 11:50, 30 March 2023 (UTC)

I tried a while ago to ensure that all of the words that are used as the names of lexical sets are in the table as examples so that someone Ctrl-F-ing will find them. I think the only one which isn't present is bath. (Are any others missing?) I am inclined to add footnotes to the ɑ and/or æ lines mentioning and briefly explaining the words affected by (and linking to the Wikipedia article on) the split. - -sche (discuss) 23:36, 27 November 2025 (UTC)

another very (should-be-)similar page

https://en.wikipedia.org/wiki/International_Phonetic_Alphabet_chart_for_English_dialects

maybe combine them, maybe make them do different things,, Swerup (talk) 11:55, 30 March 2023 (UTC)

Inconsistent rhotic 'r'

Why is the US 'car' vowel /ɑɹ/ but the 'fur' vowel /ɝ/? Should either be /ɑɹ/ and /ɜɹ/ or /ɑ˞/ and /ɝ/, depending on analysis. Since intervocalic 'r' is transcribed /ɹ/, and there is no distinction, IMO it would be hard to justify the latter, but either way we should be consistent. Currently, we have the bizarre situation where we claim RP has a consonantal /ɹ/ in a word like 'her' before a vowel, but GA does not. kwami (talk) 19:30, 9 June 2023 (UTC)

Because in "car" there's a vowel immediately followed by /ɹ/, hence notated as two separate characters, whereas in "fur" the vowel itself is R-colored, hence notated as a single composite character (and the R-colored vowel in question is /ɚ/, not /ɝ/). Whoop whoop pull up ^{Bitching Betty ⚧️ Averted crashes} 03:20, 26 June 2023 (UTC)

⟨ɚ⟩ is a reduced vowel, but fur has a full (stressed) vowel, so it's /ɝ/ in that system.

Also, this is a phonemic transcription, so whether it's /ɝ/ or /ɜɹ/ or simply /ɹ/ is a matter of analysis. Personally, I analyze it as a syllabic consonant /ɹ/, but this isn't the place to argue about theory, it's simply a pronunciation guide. It's problematic to have two conflicting transcription systems when there is no phonetic difference, and arguably no phonemic difference. kwami (talk) 04:39, 26 June 2023 (UTC)

Agree. I never know when to R-colour and when to use /ɹ/ when doing AmE. – Nixinova [‌T|C] 10:25, 25 June 2023 (UTC)

Okay, I made them consistent. See if that works for you. kwami (talk) 21:17, 25 June 2023 (UTC)

Also, /oɹ/ should not be a thing. It's two phonemes, so this notation assumes US has /o/, when it only has /oʊ/. – Nixinova [‌T|C] 06:37, 26 June 2023 (UTC)

Yes, and there's also /er/ which should be /eIr/. There are several such adjustments that need to be made, which stand out more clearly once we transcribe /r/ consistently. kwami (talk) 08:56, 26 June 2023 (UTC)

@Kwamikagami, Nixinova: there was an extensive discussion on /oɹ/ which I didn’t participate in as I’m not knowledgeable about this matter. You should probably raise any proposed changes to this page at the Beer Parlour. — Sgconlaw (talk) 11:59, 26 June 2023 (UTC)

I just stumbled upon this change and don't find it an improvement. /ɜɹ/ suggests to me a pronunciation that doesn't occur in American English - more like something that you'd hear in Scotland. I'm also surprised by Kwami's argument above that this is a phonemic transcription - it seems to be quite devoted to the details of the actual phonetic realisations in the different varieties, so this sudden turn to abstraction can have a misleading effect. For example, it suggests that the GA and RP pronunciations of a word like fur become the same as soon as there's a vowel after it (fur and..., which they certainly aren't.--87.126.21.225 04:31, 1 July 2023 (UTC)

As I said, I'd be happy with /ɹ/, which is how I analyze it. Would that be better? I was trying to change as little as possible, and just iron out the inconsistencies. kwami (talk) 04:34, 1 July 2023 (UTC)

/ɝ/ and /ɚ/

Recently, the symbols ɝ and ɚ were deleted from the GenAm section (third-to-last row and last row) of the vowel chart. They shouldn't have been deleted! If the current symbols of ɜɹ and əɹ respectively should have simply been included along with a comma rather than deleting the standard symbol. SacredForest777 (talk) 21:25, 20 July 2023 (UTC)

t͡θ

Theknightwho: I was quite surprised to see t͡θ referred to as a phonemic consonant: it is absent from any description of English consonant phonemes that I have ever seen. Is there any source for that analysis? "eighth" is certainly an unusual word, but what obstacle is there to interpreting it as ending in a cluster /tθ/, like the /dθ/ cluster in "width" or the /nθ/ cluster in "tenth" (both of which I think are words that might also have [t͡θ] on the phonetic level in my pronunciation)? The transcription /eɪtθ/ fits also with its morphological structure as eight /eɪt/ + -th /θ/. Urszag (talk) 00:09, 27 October 2023 (UTC)

I was looking at the Dictionary of the British English Spelling System, which is more interested in the fact that it's an irregular pronunciation of ⟨th⟩ (and ambiguous on whether it's /tθ/ or /t͡θ/). That being said, as a native speaker of RP British English, I'm not convinced that /tθ/ is the right analysis, since the sound is realised as a single unit, not as two separate phonemes. It's certainly possible to pronounce it as /tθ/, but it sounds really unnatural to my ear. Theknightwho (talk) 00:15, 27 October 2023 (UTC)

It's definitely an irregularity from the point of view of spelling to pronunciation. A similar irregularity is found in the spelling of "eighteen", which is not flapped in American English: it seems more parsimonious to assume an unwritten geminate /ˈeɪtˈtin/ than to postulate a new phoneme /tː/ to account for this single data point. The way that a sound is realized is a matter of phonetics, not phonemic description: it being pronounced as a single sound would justify the transcription [t͡θ], but doesn't necessarily justify the phonemic analysis /t͡θ/. This seems parallel to me to the case of words ending in /ts/: plural formation makes it clear that this is /t/ + /s/ on the phonemic level, whereas I think it is often realized phonetically as an affricate (for comparison, in German, which does have [t͡s] as a phoneme, I think word-final /ts/ and /t͡s/ are basically merged in realization). Many other phonemic clusters in English are realized phonetically as something more complicated than two phones in sequence: e.g. /tr/ and /dr/ are often affricated, /pl/ and /kl/ show devoicing of the sonorant, /ms/, /ns/, /mf/, /nθ/ etc. are prone in some accents to having epenethetic stops or affrication.--Urszag (talk) 00:37, 27 October 2023 (UTC)

I removed it from the table; I don't see a basis for viewing it as one consonant as opposed to a sequence of two. (To add to Urszag's point, /ts/ even occurs word-initially in a few words where it's a 'unit' as opposed to t + s across a morpheme boundary, but AFAIK it is still considered /ts/ even there.) If you disagree, let's (lɛt͡s?) take this to the BP and seek more input... - -sche (discuss) 02:43, 13 December 2023 (UTC)

û and ûr

Are û and ûr the same? See [2] and Zhouzhi, Zhijiang. --Geographyinitiative (talk) 18:47, 11 November 2023 (UTC)

Spaning columns stops easy look-up by eye

Because some table cells span multiple columns, e.g. eɪ, they're not found when scanning down a column. This is annoying as there are normally multiple things to look up and memorise as the whole word's pronunciation is attempted. The spanning cells should be replaced with ones which cover just a 1x1 space in the table even though that means duplication. Ralph Corderoy (talk) 14:08, 9 December 2023 (UTC)

Cannot decode ɛ in /ˌɛdəˈmɑːmeɪ/

There is no entry for a lone ɛ. Ralph Corderoy (talk) 13:44, 13 January 2024 (UTC)

@Ralph Corderoy: yes there is—see row 7 of the vowel table. — Sgconlaw (talk) 23:18, 4 April 2024 (UTC)

Thank you, @Sgconlaw. I see it now. Ralph Corderoy (talk) 12:17, 16 April 2024 (UTC)

Introducing awful and awning as examples of ä

I have a suggestion. If my suggestion is not implemented, then the table will look like this...

enPR / AHD	IPA					examples
	RP	GenAm	CanE	AuE	NZE
ä	ɑː	ɑ	ɒ, ɑ	ɐː, aː		father, palm

---

If my suggestion is implemented, then the table will look like this...

enPR / AHD	IPA					examples
	RP	GenAm	CanE	AuE	NZE
ä	ɑː	ɑ	ɒ, ɑ	ɐː, aː		father, palm, awning, awful

---

Note that almost every English word containing aw has the aw pronounced as ä

Thus, we can provide the following examples: awful, awning.

A hawk atop some straw above the awning produced an awful squawk when the fawn walked on the lawn below.

A häk ätäp some strä above the äning produced an äful squäk when the fän wäked än the län below. 2601:280:C081:8A50:C95F:7337:3DA4:9B47 23:10, 4 April 2024 (UTC)

No. "Awful" only merges with the "father, palm" vowel for speakers who have the cot-caught merger; this is reflected later, in the "law, caught, thought" line. - -sche (discuss) 23:39, 4 April 2024 (UTC)

Tables for vowels and constonants disagree on column ordering

The vowels and foreign vowels tables have IPA second whereas consonants has it first. This is tedious when skipping up and down between the tables to eke out something like niːtʃə. The three should be consistent. Ralph Corderoy (talk) 13:34, 12 May 2024 (UTC)

start borrowing horses

Re diff, what I was getting at is that our "enPR" simultaneously uses different symbols for the same sound, and uses the same symbol for different sounds, which seems confusing. The table, as currently laid out, seems to suggest that "ŏr" be used to mean both "oɹ" and "ɑɹ" (as in borrow), but that oɹ in horse be "ôr", and ɑɹ be "är" (as in start). I am asking whether this is the best idea, or whether (when a word can be pronounced either oɹ or ɑɹ) we should just represent it as having both oɹ or ɑɹ (and thus, in enPR terms, both ôr and är), so that one symbol stands for one sound, consistently wherever that sound occurs. In other cases where a word sometimes has one sound and sometimes another, we notate the sounds separately (e.g. /ɪ/~ĭ vs /aɪ/~ī in privacy, or /ɛ/ vs /i/ in decal). - -sche (discuss) 22:50, 16 September 2024 (UTC)

borrow, sorry, sorrow, (to)morrow are the only words that have this distribution. I suggest removing the row altogether. And the enPR for /oɹ/ is always ôr: [3]. ŏr should only be used for /ɑɹ/ spelled or, which is a dialectal minority except for these five words. Nardog (talk) 02:39, 17 September 2024 (UTC)

I was about to agree with removing the two ŏr rows, and dispersing the class of words which have this distribution (sorrow, orange, horror, etc) into the other relevant rows... but then I realized: the ŏr rows are needed for RP. Though GA only has /ɑɹ/ and /oɹ/, RP has /ɒɹ/ as a third sound contrasting with both RP /ɑːɹ/ (starry) and RP /ɔːɹ/ (forum). Hmm... can we edit the GA cells so that they explain that ~"words with RP /ɒɹ/ (ŏr) have /oɹ/ (ôr) or /ɑɹ/ (är) in GA, and the GA pronunciation should use those symbols"? After all, we put separate RP vs GA enPR lines on other entries where RP and GA have different sounds, like privacy... the one(?) other place we're using two symbols for one sound is when we use both ŏ and ä for GA /ɑ/, motivated by RP having different sounds, but we might want to rethink that too... - -sche (discuss) 01:45, 18 September 2024 (UTC)

Do we use enPR for anything other than GA? I don't see reason to. Nardog (talk) 11:07, 18 September 2024 (UTC)

@Nardog We sometimes add it to RP pronunciations, but we probably shouldn't, since the vowel system doesn't really fit. Theknightwho (talk) 11:48, 30 September 2024 (UTC)

Yeah, people seem to use "enPR" as pan-dialectal / diaphonemic notation... which is, I suppose, why the table has these lines, to let one symbol indicate both the RP and GA pronunciations. But one result of that is the awkwardness we're discussing. But even if we 'fix' that awkwardness, in any of the ways I've suggested, or even by saying enPR should no longer be used for RP (a comparatively more significant change, in terms of how many entries it would effect), it occurs to me that we'll be facing this same question (whether to let one input stand for two sounds, and in which cases, and how) in the future if any {{en-IPA}} project—yours, Benwing's or anyone else's—expects to take one input and output both RP/UK and GA/US, won't we? The more I think about this and the more we talk about it, the less sure I am of what to do. :/
I suppose one 'principle-of-least-change' way we could resolve the awkwardness of "one symbol stands for two sounds" without entirely removing the line would be to define enPR "ŏr" as meaning, in GA, only /oɹ/ or only /ɑɹ/, and say that whenever a word is instead/also pronounced in GA with the other of those sounds, the usual symbol for that sound should be used instead of "ŏr"...? Or (as mentioned above) define "ŏr" as something that should only be used for RP and other dialects, with GA using the different symbols that represent the different sounds GA has? (Like the enPR pronunciations of RP vs GA privacy, and presumably any {{en-IPA}} inputs that generate them, will also need to differ...) - -sche (discuss) 19:54, 30 September 2024 (UTC)

Another possible edge-case: /t͡s/

@-sche @Urszag Is /t͡s/ another possible edge-case phoneme that could go under the consonant section? This one pretty much only occurs in borrowings, but definitely crops up in my speech when I say Mozart, intermezzo, pizzicato, sforzando. Particularly with sforzando, the stress falling on the second syllable leads to [sfɔːˈt͡sændəʊ], which is noticeably distinct from outside (adverb) [aʊtˈsaɪd] due to the co-articulation. If you have access, compare the OED's recordings for the two words, as the difference is clear. The question is whether this is phonemic, I suppose, but I can't think of any minimal pairs. Theknightwho (talk) 12:21, 30 September 2024 (UTC)

The pronunciation that you describe can be interpreted as /ts/ pronounced in one syllable (whereas "outside" contains /ts/ pronounced in separate syllables), which is not necessarily a single phoneme any more than /st/ or /tr/ is (both of these also sound noticeably different when pronounced at the start of a single syllable vs. split across two separate syllables). While you could call it an affricate by analogy with /t͡ʃ/ /d͡ʒ/, or based on it being a phoneme in the source language, this is not necessary. Plus, in cases where the stress is not on the following vowel, I doubt there is a distinction from native English /ts/, as in "moats", "nets", "pits".--Urszag (talk) 12:49, 30 September 2024 (UTC)

@Urszag It hinges on whether it can be used syllable-initially, which is not typically allowed in English. That's why I distinguished sforzando and outside. The same restriction applies to /t͡ʃ/ and /d͡ʒ/, where non-coarticulated /tʃ/ and /dʒ/ cannot occur syllable-initially either, which is why the co-articulated forms are analysed as phonemic in the first place. Theknightwho (talk) 13:13, 30 September 2024 (UTC)

If by "it" you mean "whether it is a phoneme", I don't agree that this necessarily follows. There's no question that "ts" is not typically allowed at the start of a syllable (so using it in this position is a foreignism), but we also may find /ps/, /ks/, /kʃ/, /pθ/ at the start of a syllable in some foreignisms, and nobody would consider those phonemes. Coarticulation is a phonetic, not a phonological criterion: the fact that [ts] can phonetically be characterized as an affricate does not imply that it has the phonemic status of a single consonant. Phonemic analyses are famously non-unique, so I don't mean to say it's impossible to treat /ts/ as a phoneme in English, but the facts don't mandate it, and if you think Wiktionary should do so I think you should look for some precedent for that treatment. I imagine some dictionaries might have made that choice, but I haven't seen it.--Urszag (talk) 13:25, 30 September 2024 (UTC)

@Urszag How would you distinguish /t͡ʃ/ from /tʃ/? Theknightwho (talk) 13:39, 30 September 2024 (UTC)

English doesn't have any contrastive distinctions between affricates and stop-fricative sequences within the bounds of a single syllable, so in principle, it is only necessary to mark syllable boundaries when those are not obvious, as in courtship, nightshade. Hence why many phonologists, among them John Wells, don't bother to use tie bars on affricates in English (e.g. Wells' Longman Pronunciation Dictionary transcribes chip as /tʃɪp/—which should not be understood as a claim that this word is made up of four phonemes /t/, /ʃ/, /ɪ/, /p/). In terms of phonology, /t͡ʃ/ is analyzed as a single consonant because non-foreignisms do not otherwise start with stop-fricative sequences, among other reasons. Wells writes on affricates "English has two affricate phonemes: tʃ, as in church tʃɜːtʃ || tʃɝːtʃ, and dʒ, as in judge dʒʌdʒ. [...] we do not usually list bv, tθ, ts, dz among the English affricates. Notice also that the t followed by ʃ in nutshell ˈnʌt ʃel is not an affricate." (LPD, 2nd ed., p. 15) I think Wells' position is pretty standard but if you can find a phonologist who analyzes English differently, that might be worth mentioning on this page.--Urszag (talk) 14:09, 30 September 2024 (UTC)

@Urszag Well, the problem with that argument is that it entirely hinges on where you draw the syllable boundary, and I'm not sure that it's reasonable to draw a distinction between native terms and borrowings, because that doesn't explain what makes /t͡ʃ/ phonemic. This leaves us with a problem:

1. Either we accept that the distinction between /t͡ʃ/ and /tʃ/ is phonemic because we consider that being affricated in a phonemic distinction in English (e.g. nutshell vs *nutch-ell), which leads to exactly the same conclusion when applied to sforzando (affricated /t͡s/) as compared to outside (non-affricated /ts/). Again, the difference occurs due to compounding.
2. Or, we have to say that affrication is not phonemic, in which case we shouldn't be using /t͡ʃ/ at all in phonemic transcriptions.

I'm not really satisfied with the approach of "just follow what others do", because I'm looking for some consistency in our analysis here. Since this isn't WP, we aren't totally constrained by what the sources say. Theknightwho (talk) 14:36, 30 September 2024 (UTC)

A phoneme is a unit of phonology. We can see that if we add the genitive suffix -'s to the name Walt, we get the form [wɔlts], which implies that syllable-final [ts] should be analyzed as a sequence of two consonant phonemes, /t/ and /s/, in at least some cases. Since the unitary morpheme waltz is pronounced the same way, it is simplest to give both Walt's and waltz the same phonemic representation, rather than positing an inaudible phonological distinction between /wɔlts/ with underlying /t/ + /s/ and /wɔlt͡s/ with an underlying affricate /t͡s/. In contrast, it’s difficult to come up with examples of words with a morphological structure that requires analyzing tautosyllabic /tʃ/ or /dʒ/ in English as a sequence of two separate phonological units. So treating /t͡s/ as a phoneme causes problems that don’t arise from the common practice of taking /t͡ʃ/ /d͡ʒ/ as phonemes in English. Even if we aren't constrained by what reliable sources say, I'm satisfied with the mainstream phonological treatment of this topic, and I think that when dealing with the phonological analysis of a highly documented languages such as English, the burden lies on proponents of novel analyses or symbols to demonstrate why they are necessary. It isn't like we are the first people to have considered these issues.--Urszag (talk) 15:39, 30 September 2024 (UTC)

Why is the silent R in non-rhotic accents not indicated?

In RP, in a word like customer or boar, there is an underlying /r/ which is silent unless the word is followed by another word beginning with a consonant (linking R). Why is this not indicated? Compare Italian, where words that trigger syntactic gemination are indicated by an asterisk, e.g. più. In some varieties of British English, an "intrusive R" is inserted even when not etymologically present; these varieties can be analysed as having no underlying /r/. But the term "RP" would seem to indicate that a conservative variety without intrusive R is being used.

Cambridge Dictionary uses a superscript ⟨r⟩ to indicate an underlying /r/. Oxford Learner's Dictionaries uses an ⟨r⟩ in parentheses. Un assiolo (talk) 15:22, 25 October 2024 (UTC)

As I understand it, we don't write underlying rs because people who distinguish them are rare even among older speakers of old-fashioned RP in television and so on. Most people use the so-called intrusive or linking r. There isn't a rule against adding transcriptions that indicate an underlying r, but they should be marked as historical transcriptions, like transcriptions showing the early 20th-century distinction between north and force, or the even older distinction between bark and Bach. — Eru·tuon 17:21, 28 December 2024 (UTC)

I agree that linking R is basically the norm in whatever can be described as modern RP, as John Wells says. The designation RP is not, I believe, intended to describe a particularly old-fashioned upper-class accent - it is just used for lack of a commonly accepted standard alternative to refer to the middle-class South-Eastern British variety that is perceived as generic British English and which Geoff Lindsey calls 'Standard British'. In practice, as of now, the transcriptions in some entries do use /r/ in parentheses for word-final positions, even though others (perhaps most) don't. Even worse, this usage is not explained on this page, so people can get the wrong impression that these Rs are simply 'optional', in the sense that some RP speakers pronounce them and some don't. And indeed, some editors seem to have got more or less this impression and just write them in parentheses even before consonants, where they can never be realised in RP - while often also writing 'UK' instead of 'RP', thereby doing, in practice, some kind of interdialectal transcription compromising between non-rhotic and rhotic accents in the British Isles.--62.73.72.3 07:18, 31 December 2024 (UTC)

Rhoticity in Indian English

The transcription of Indian English in the examples on the page is very inconsistent in terms of rhoticity - a rhotic pronunciation is shown after some vowels and a non-rhotic one after others. I don't think that there is just a dependency in reality - rather, this is just accidental chaos due to inattention. As for what the consistent transcription should be - since some Indian English speakers are non-rhotic (not always consistently) and others are rhotic, arguably the /r/ should be in parentheses everywhere. 62.73.72.3 07:33, 31 December 2024 (UTC)

RP /ɛː/ and /ɵː/?

The footnote says that "RP in the early 20th century had five centring diphthongs /ɑə/, /eə/, /ɪə/, /ɔə/, /ʊə/. ... All of them are now generally pronounced as long monophthongs (pure vowels) /ɑː/, /ɔː/, /ɛː/, /ɪː/, /ɵː/." However, while /ɑː/, /ɔː/, and /ɪː/ all show in the RP column, /ɛː/ and /ɵː/ are nowhere to be found. Shouldn't they be either added to its column or removed from the footnote? 209.195.249.6 15:25, 3 February 2025 (UTC)

We're giving the common phonemic transcriptions here, which may not match the precise phonetic realizations. The traditional phonemic transcriptions for the NEAR, SQUARE and CURE vowels are /ɪə/, /eə/ and /ʊə/, respectively. However the symbol /ɛː/ is certainly widely used in phonetic transcriptions of RP, for example by the Oxford English Dictionary [4] [5] amongst other sources, so if we're listing /a/ alongside /æ/ for TRAP (another OED deviation from the traditional symbols, less widely used than /ɛː/), and /ɪː/ for NEAR alongside /ɪə/ (which is much less common, only used in Geoff Lindsey's SSB transcription that I know of), we should certainly list /ɛː/, which we in any case already use on many entry's Pronunciation pages, such as laird, so I'll add it. Not to mention that

Re /ɵː/, it's only used in Lindsey's SSB transcription, which is explicitly defined as not an RP transcription system, so arguably if we're listing /ɪː/ we should list /ɵː/, but I'll leave it off for now since we don't list most of Lindsey's SSB symbols and I don't think we use it on any entry pronunciation pages.

I'll change the footnote to show that the monophthongs are phonetic transcriptions, not necessarily intended to match up to the phonemic transcriptions. Offa29 (talk) 23:58, 3 February 2025 (UTC)

/ɔ/ for GA LOT

Watchers of this page are invited to review Special:History/pot and Special:History/shot. Nardog (talk) 17:48, 16 February 2025 (UTC)

Or Wiktionary:Tea room/2025/February#pot, shot. Nardog (talk) 18:11, 16 February 2025 (UTC)

GA vs US label in the template

Some weird technical change has been made causing the GA templates to display as 'US' instead. For example, in the entry for grasp, the syntax has 'GA', but what the reader actually sees is 'US': (General American) IPA^(key): /ɡɹæsp/. This is totally misguided, since the US has many different accents, whereas GA is just one of them. Another reason why it is inappropriate is that this appendix page, which all the transcriptions link to in order to help the readers interpret them, describes GA and not 'the US accent'. I don't know how to undo this change to the template, but it definitely should be undone. Anonymous44 (talk) 14:19, 22 March 2025 (UTC)

@Anonymous44: I agree this is wrong. @Theknightwho, any idea why this is happening? — Sgconlaw (talk) 17:32, 22 March 2025 (UTC)

@Sgconlaw It's because @Jaiganesh.kumaran merged a bunch of labels in Module:labels/data/lang/en. I have reverted that change for now, as it needs discussion. Theknightwho (talk) 17:40, 22 March 2025 (UTC)

@Theknightwho: thanks! I noticed those changes and reverted a major one, but didn’t know “Module:labels” was linked to {{Module:IPA}}. — Sgconlaw (talk) 17:55, 22 March 2025 (UTC)

Apologise for the silent change then, altho there is a lot of inconsistency in how the accent labels are used and the phonemic system they use. I'll discuss more about this later. Jaiganesh.kumaran (talk) 20:13, 22 April 2025 (UTC)

Some things that I noticed, that could be improved

"English" (beginning): Scottish English (ScE) is not mentioned at the beginning ("The sounds of [...] are shown."), even though the "Vowels" table includes it.

"Vowels" (in the table, ô [EDIT: now also ŏ], GenAm): I think a ref tag should be used to separate the IPA from the note, for example:

{{IPAchar|ɔ}}, {{IPAchar|ɑ}}<ref>{{IPAchar|ɑ}} for Americans with the [[w:cot-caught merger|''cot''–''caught'' merger]], {{IPAchar|ɔ}} for Americans without it.</ref>

"Vowels" (at the end, before "Consonants"):

• Since the disyllabic sequence /iə/ is mentioned, I think what should also be mentioned is, is that it should be replaced with /iə̯/ (see ◌̯) if actually transcribing âr (alt: /eə̯/) or îr (alt: /iːə̯/) in NZE. And possibly add the breves to the IPA table as well, just in case.
• It seems the bath vowel is ä, split from the trap vowel ă, so the ä examples should include bath and mention the split with a ref tag; see also, the trap–bath split.

"Other symbols": the note seems to be mostly out of date, online AHD seems to be using weird Unicode characters (the PUAs or w/e) to distinguish between primary and secondary stress, see battleship as an example. I said "mostly", because I just found the entry strong-minded, that uses this old system where you can't differenciate the stress. (or maybe both words are stressed primary, as if they were separate?) Unrelated: I also found an error in type, where the pronunciation is written as *tip (and *i doesn't exist on its own, only in oi, it seems) instead of tīp.

Thank you, 83.28.247.254 17:00, 19 April 2025 (UTC)

Replace phoneme tables with sectioned text

The phonemes table looks too messy now. I'd prefer listing the phonemes in text one by one, considering that we only document 7 accents. Jaiganesh.kumaran (talk) 19:28, 22 April 2025 (UTC)

The table looks fine to me, and a table is by far the most suitable and easily navigable format for this kind of material. Also, seven accents isn't a small number.--Anonymous44 (talk) 10:36, 20 May 2025 (UTC)

STRUT-COMMA merger in General American varieties of English

The STRUT lexical set and the COMMA lexical set are merged in most American varieties of English, especially those considered 'General'. The practice of representing the phoneme in broad transcription solely as schwa has been adopted by progressive dictionaries such as American dictionary Merriam-Webster and British (Oxford) dictionary Lexico.

As John Wells describes in his books Accents of English 1 and Accents of English 3, 'in [General American varities] it may well be considered that stressed [ʌ] and unstressed [ə] are co-allophones of one phoneme' (1, p. 132) and that '[s]ince [General American varieties...] usually lacks a proper opposition between [ʌ] and [ə], it follows that the phoneme may equally be written /ə/' (3, p. 480).

It should be the case that on the vowel page the vowel represented by enRP ŭ should list its pronunciation in General American pronunciation, in the best case scenario, only as ə or, if not, at least both ʌ and ə.

The same is true for Standard Canadian English.

, Summer9778 (talk) 16:56, 7 August 2025 (UTC)

those are two different sounds. it's /ʌ/, not /ə/. there's a difference between omnibus with a full vowel and a reduced vowel, at least for many speakers of GA. kwami (talk) 23:33, 7 August 2025 (UTC)

This has been discussed before a few times, e.g. Wiktionary:Beer parlour/2022/November#ʌ in American English pronunciations and Wiktionary talk:Votes/2023-12/Represent the GenAm NURSE and STRUT vowels as schwa. I have not gotten the impression that there was clear consensus/support for making this change to our notation yet; some people want to make the change and some people don't. At the moment we mention this in a footnote. - -sche (discuss) 04:37, 8 August 2025 (UTC)

General Australian

Page-watchers, please see Wiktionary:Tea room/2025/August#AU should be Australia. - -sche (discuss) 05:12, 16 August 2025 (UTC)

Category and label treatment request: August 2025

The following discussion has been moved from the page Wiktionary:Category and label treatment requests (permalink).

This discussion is no longer live and is left here as an archive. Please do not modify this conversation, but feel free to discuss its conclusions.

---

bowl-bull merger

for English, I request the accent label |a=bowl-bull for the bowl-bull merger. see example at you'll. Juwan (talk) 13:53, 19 August 2025 (UTC)

Hmm, that entry (you'll) lists the merger outcome as /ˈjoʊl/, but in the past, people who've had this merger (and thought and insisted everyone in the US had this merger) have said the outcome (for bowl-bull) is /bl̩/ (so, for you'll, /jl̩/, not distinguishable from the unstressed pronunciation, /jəl/). My initial reaction is that we may need this phenomenon to be better documented before we can include it at all. - -sche (discuss) 15:22, 19 August 2025 (UTC)

@-sche good catch. I didn't properly look into this one. Juwan (talk) 15:57, 19 August 2025 (UTC)

It seems like a handful of /oʊl/ pronunciations have been added by one(?) user, and may need to be verified (or else removed). It's conceivable that different regions have different merger outcomes (WP suggests that in the UK the merger results in the loss of the /l/, whereas the various merger pronunciations I've seen Americans claim have all kept the /l/), but IIRC the user saying the outcome was /bl̩/ was also Californian, whereas these entries claim the Californian outcome is /boʊl/. (And here, one user says they've heard diphthongized bull as the way they merge, whereas another says they monophthongize bowl, to "something like [ɤ]". Interestingly, another user says that a following vowel blocks the merger, so bully and holy and gully remain distinct.) What I found when I looked into this before is the same thing I find now, which is : a few brief mentions in reference works that some American speakers (from the PNW, from the Midwest, etc; different works discuss different regions) merge some or all of /ʊl/, /oʊl/, /ʌl/, and /ɔl/ (often citing Labov), every one of which unfortunately fails to say what the merger outcome is; I don't think we can include it (and so, we would need to remove it from bull, full, pull, unfull, wool) unless we can verify in reference works what the merger pronunciation is. - -sche (discuss) 17:30, 19 August 2025 (UTC)

I have left a comment on the talk page of the user who added these in the hope that they know of reference works documenting the merger outcome. The previous major discussions of this were Wiktionary:Tea room/2014/September and Wiktionary:Tea room/2015/July (Ctrl-F "bull", because for some reason generating a section link on archive pages is difficult; I will raise that in the GP). - -sche (discuss) 18:02, 19 August 2025 (UTC)

I've removed the merger pronunciation from the 6 entries which included it, given that we can't confirm the (merger outcome) pronunciation; indeed, when I search again now for any reference works documenting the merger and its outcome, I find this which mentions a version of the merger that also involves /u/, "The merger of bull, bowl, and boule is a Western vowel system feature with a specific phonetic distribution: preceding lateral /l/"; it implies that a merger of /ɪ/ with (to?) /i/ before /l/ may be the same underlying phenomenon, "the prelateral merger (feel/fill, bowl/bull), which can be associated with a raising of both vowels to the quality of the tense member of each pair"; and crucially, it suggests that unlike e.g. the cot-caught merger where the outcome is straightforward, "/ɔ/ → [ɑ]", the outcome of the bull-bowl-boule merger is variable, "/ul/ ⟷ /ol/, /ʊl/ ⟷ /ul/, /ʌl/ ⟷ /ol/".

Other data confirms that the outcome of merging bull and/or bowl and/or boule varies between speakers, even in the same families in the same cities.

This discusses the merger in Kansas City, and says: "Impressionistically, words like bull and pull are pronounced with something close to [o], so that they are occasionally homophonous with bowl and pole. This conditional merger was not explored in ANAE, which instead explored the conditional merger of GOOSE and FOOT before /l/ (e.g., pool and pull—cf. Ash 2006:42-43 for notes on pre-/l/ FOOT in ANAE)." The paper complains that the dataset it's discussing would've retained examples where someone said "pool as [pʊl] and even bull as [bul]" but would've thrown out, and thus prevented the current paper's author from measuring, "bull as [bol]". This, also about Kansas City, says: "Results show that BULL frequently merges with either POOL or BOWL in Kansas City, though there is little evidence of speakers merging all three vowels together. Moreover, the two mergers behave differently from one another in apparent time in terms of both production and perception. Apparent-time evidence indicates that the merger of BULL and BOWL has been spreading gradually in terms of production, but took hold rapidly (in a single generation) in terms of perception. By contrast, the perception of BULL and POOL as merged has spread gradually, but their production as merged has actually retreated in apparent time. Acoustic analyses suggest that the relevant changes are in the F2 of POOL and BOWL—BULL appears to have reached a very back production in Kansas City at an earlier stage in the development of the city’s dialect, but continued backing of POOL and BOWL among younger Kansas Citians is pushing POOL away from merger with BULL and drawing BOWL into merger with BULL. To seek explanations for the different patterns of merger, the paper concludes by examining case studies of individual interviewees. These include, for instance, two sisters born in the 1990s, one of whom is BULL-POOL merged and the other BULL-BOWL merged."

Hopefully more people study the merger and document its outcomes in different places. - -sche (discuss) 19:53, 24 August 2025 (UTC)

I have added links to the section title so that when this is archived, links are hopefully left on relevant pages. - -sche (discuss) 04:31, 26 August 2025 (UTC)

The 'Mary' vowel in General American

Entries like wear and bear currently have /ɛəɹ/ in the GA transcriptions. However, the table indicates that it should be /eɹ/ (or, in merged accents, /ɛɹ/, but they are predictable based on the unmerged ones and hence don't need to be rendered). Seems like many entries should be changed, or else the table should be. ~2025-33239-81 (talk) 18:18, 23 November 2025 (UTC)

Alas, the failure of Wiktionary's many unrelated volunteers and drive-by editors to standardize on either /ɛəɹ/ or /ɛɹ/ (and similarly /ɪəɹ/ or /ɪɹ/, etc) when adding pronunciations is longstanding and quite possibly unavoidable. (The merger pronunciations are listed because most Americans / GenAm speakers have the merger.) - -sche (discuss) 22:45, 7 December 2025 (UTC)

lack of 1-to-1 relation between enPR and sounds

See Wiktionary:Beer parlour/2025/December#lack of 1-to-1 relation between enPR and sounds. - -sche (discuss) 22:44, 7 December 2025 (UTC)

Indian English: Use of /əɹ/ or /ɚ/

@Jaiganesh.kumaran: regarding your recent edits, if we are using /əɹ/ in other varieties of English (there was a previous discussion on this issue), then for consistency it seems to me we should do the same for Indian English rather than use /ɚ/. (Pinging @Kwamikagami, Mahagaja, -sche, Theknightwho in case they wish to comment as well, as I'm not an expert in this area.) — Sgconlaw (talk) 12:32, 8 December 2025 (UTC)

I'm just treating /ɚ/ as equal to /ə(ː)(r)/ - in Indian English stressable NURSE is often interchangable with unstressed lettER

egː colorful can be /ˈkələ(r)fʊl/ or /kəˈlɜ(r)fʊl/ - which I plan to just transcribe as /kalɚful/ without stress markers and use the module en-Indic I'm developing to generate dialect-specific transcriptions that only include either one or both.

I am currently in the process of cleaning up the transcriptions so I may change my mind soon; there is lots of inconsistencies in the notation used on different entires. Sorry for changing things in secrecy tho. Jaiganesh.kumaran (talk) 12:57, 8 December 2025 (UTC)

I have reverted this for now. The alternative is listing ə(r) and ɜ(r) separately as done before. Jaiganesh.kumaran (talk) 13:13, 8 December 2025 (UTC)

@Jaiganesh.kumaran: no problem. Just thought it's better that we be consistent, otherwise it's rather confusing. — Sgconlaw (talk) 13:24, 8 December 2025 (UTC)

Although there are several GA entires using ɚ over əɹ, there is no enforcement of consistency on this siteǃ Jaiganesh.kumaran (talk) 13:26, 8 December 2025 (UTC)

@Jaiganesh.kumaran: the problem was that there was a discussion with, apparently, a consensus to adopt /əɹ/ and /ɜɹ/ over /ɚ/ and /ɝ/, but no subsequent effort to replace the uses across the entire site using a bot (maybe it wasn't advisable for some reason). Thus, it's done manually by editors as and when they come across such entries. — Sgconlaw (talk) 13:33, 8 December 2025 (UTC)

In GA what is phonemically /ər/ is realized as [ɚ]. Nardog (talk) 13:45, 8 December 2025 (UTC)

Those are not equivalent. ⟨ɚ⟩ represents one segment, ⟨əɹ⟩ represents a sequence of two. See Handbook of the IPA, p. 25. Nardog (talk) 13:42, 8 December 2025 (UTC)

@Nardog: I can't comment on this, but would just point out again that there was a discussion about this which led to the wholesale replacement of /ɚ/ and /ɝ/ with /əɹ/ and /ɜɹ/ in this appendix. If you think that was inadvisable, then a fresh discussion should be started at the Beer Parlour, hopefully involving as many knowledgeable editors as possible. It isn't desirable for us to keep flip-flopping on this issue because it's confusing for editors like me. — Sgconlaw (talk) 14:30, 8 December 2025 (UTC)

I don't think that was inadvisable. /ər, ɜr/ are the more theoretically sound analysis if you ask virtually any phonologist. Nardog (talk) 14:37, 8 December 2025 (UTC)

For the record, I don't know whether or not there is a consensus about which notation is better: it's been discussed a number of times (Wiktionary:Beer parlour/2024/February#Changes from /ə(ɹ)/ or /əɹ/ to /ɚ/ might be the most recent) and I'm unsure if there's been a clear resolution / decision, but I agree it's best to use one notation consistently, and I do opine that consistently using /əɹ/ has the advantage of being consistent with our notation of other vowels — we don't write start as /ɑ˞/ or north as /ɔ˞/. - -sche (discuss) 17:32, 10 December 2025 (UTC)

Edit request

For examples for the ō vowel (non-toe-tow merger section), please add a link to {{l|en|t'''ow'''}} in the fourth place, so the list would be {{l|en|kn'''ow'''}}, {{l|en|s'''ou'''l}}, {{l|en|r'''o'''ll}}, {{l|en|t'''ow'''}}, {{l|en|c'''o'''ld}}, which would make it visually similar to what was done for the ā vowel. Thank you, ~2025-33146-07 (talk) 09:48, 20 December 2025 (UTC)

Duplicate anchor ID: vowel in the word fur appears thrice

Why is this even a thing here? Aren't the North American er and interchangeableVR (both group lower-alpha) references enough? ~2026-72188-4 (talk) 16:37, 11 March 2026 (UTC)

Too many variants in IPA cells

E.g. for GA here, the table gives both an option with a schwa and an option without one. These aren't really different pronunciations, just different analyses/interpretations found in the literature. If the purpose of the IPA cells were just to tell people how to interpret enPR, it would have been OK to include several different transcriptions of the same pronunciation. But many entries give only IPA, and the link to this page is supposed to help people interpret the IPA by means of the examples. We should use a single standard way of transcribing a certain pronunciation, even if there are alternative justifiable options, in order to make it easy for the reader to interpret our transcriptions. And this table should indicate which this single standard way of transcribing is. Anonymous44 (talk) 16:14, 16 March 2026 (UTC)

The triphthong in hire

... is omitted in the table. It should be included. GA could be transcribed with or without one, but RP does have a triphthong. Anonymous44 (talk) 16:31, 16 March 2026 (UTC)

Interesting point; the table doesn't have separate lines for the British triphthongs in flour, power and coir, loir either, nor the marginal(?) American diphthong /iə/ (Merriam-Webster specifically acknowledges one-syllable /iə/ as a thing that exists distinct from two-syllable /i.ə/, in e.g. the two- vs three-syllable pronunciations of idea), and other such things. To my understanding, all of these are often analysed as sequences of two phonemes, and that's probably what the table is doing, and that may be alright(?). (Certainly it's not inherently a problem to have two phonemes in one syllable: the table doesn't have sequences like /st/, /sk/, /sp/ either.) Hopefully more people will weigh in if they think we should (or shouldn't) have separate lines for these. - -sche (discuss) 06:51, 16 June 2026 (UTC)

'South Asian IPA'?!

This sounds as if it were a different transcription system, some kind of alternative version of the International Phonetic Alphabet used only in South Asia. What the editor meant is presumably that the corresponding diaphonemes are pronounced differently in Indian English. So the article should just say that. Now it's a total mess. Anonymous44 (talk) 01:17, 18 March 2026 (UTC)

'UK' and 'US' labels

Wiktionary's English pronunciation entries are full of pronunciations labelled 'UK' and 'US' instead of RP and GA. This is clearly inadequate, since there are many UK and US accents, not just one, and it's clear that in practice, RP and GA are meant. Furthermore, the transcriptions link to this page as a manual for their interpretation, and this page has rows for RP and GA, not for 'UK' and 'US'. To make things worse, some of the 'UK' transcriptions seem to be genuinely trying to be 'interdialectal', but only with respect to one of a myriad dialectal variables, namely by showing rhoticity as optional (perhaps to humour those non-rhotic speakers who, under the influence of orthography and of General American, insist that there is somehow an /r/ in their pronunciations, as well as the majority of foreign language learners and teachers who pronounce rhotically while inexplicably believing that they speak 'the King's English'). Of course, this is an arbitrary choice that just shows the hopelessness of trying to make a 'pan-UK' transcription, especially using IPA. Somebody has just been following a different concept for the transcriptions than the one envisaged by this page (perhaps Wikipedia's rather misguided interdialectal/diaphonemic IPA system), possibly without realising the incongruity. Anyway, I fix these where I encounter them, but there is a myriad of them all across Wiktionary and this is the kind of process that I would expect to lend itself to automated execution by a bot. (Of course, the 'UK' and 'US' labels can be appropriate when serving merely as subheadings introducing the transcriptions of multiple specific UK dialects, as they do in some entries; or for audio files, when no more specific identification of the accent is given.) Anonymous44 (talk) 10:43, 15 June 2026 (UTC)

I don't think it's really a problem. As you say it's clear what is meant in practice. Most of the time the dictionary is just trying to express the phonemic units that distinctively make up a word, which are mostly identical across dialects, not trying to provide a dialectological compilation or comparison. I actually much prefer UK and US for general use specifically because they are not labels of outdated dialects presumed or perceived as contemporary Hftf (talk) 19:17, 16 June 2026 (UTC)

FWIW, this is a longstanding issue (it was already something that had been being discussed for a long time more than a decade ago). The large number of uses of "US" and "UK" and low number of volunteer-hours available to change them, combined with the difficulty of getting people to agree on what changes are desirable, means things are still as inconsistent ("UK" here, "RP" there, one entry with "US", another with "GA", etc) as ever. (See also: my watchlist is continually full of at least one editor [Geographyinitiative] going around adding enPR to entries and at least one editor [Orrigarmi] going around removing enPR from entries, such that I don't know if the overall number of entries with enPR is changing at all.) - -sche (discuss) 21:24, 16 June 2026 (UTC)