Wiktionary talk:Votes/2023-12/Represent the GenAm NURSE and STRUT vowels as schwa

“represent[ing] these vowels as schwa for British English”

I don't feel qualified to judge regarding General American English, whereas I do regarding British English. I didn't see anywhere good in Wiktionary:Beer parlour/2023/October § English IPA vowels to throw in my tupp'orth, so I'm writing this here.
Re point 3 of the proposal (viꝫ “To be determined before starting this vote: should we also represent these vowels as schwa for British English, like the OED does for nurse?”), the pairs /ə/–/ʌ/ and /ə/–/ɜ(ː)/ are both contrasting to my mouth and ears, so I would oppose representing either or both of the STRUT and NURSE vowels as schwa for British English.
0DF (talk) 01:03, 29 December 2023 (UTC)

I agree. Note: @-sche: The OED does use /ʌ/ for the vowel in "run" for BrE. AG202 (talk) 03:05, 29 December 2023 (UTC)

Interesting: perusing various entries, it seems like the OED takes the view that British English has merged the NURSE vowel, but not the STRUT vowel. I must say that although I set up this vote (because the issue has kept coming up and it seems like it would be good to therefore actually have a formal decision), I think I may vote oppose, because it seems like we're not likely to manage to actually update our notation of the sounds overall / for all dialects, so as long as we're sticking with the historical symbols in some cases — despite e.g. Geoff Lindsey's arguments and data that British speakers don't actually reliably produce distinctions, and can't reliably identify the isolated sounds, or that American speakers don't — I think it's simpler to stick with them in all cases. (This reminds me of how various British contributors wanted to update the notation of the TRAP/CAT vowel to /a/, like the OED does, but enough people argued that unless we also do that for GenAm, we shouldn't do it for British.)
(I've taken this stance in prior BP discussions, but if people think I should just withdraw the vote if I intend to vote against it, I'm willing to.) - -sche (discuss) 00:36, 30 December 2023 (UTC)

@-sche: I don't really understand the logic that we can't change the notation for one dialect without the changing the notation for another. Aren't they just separate issues? 0DF (talk) 01:26, 30 December 2023 (UTC)

Someone who sees a "UK" transcription like /kʌt/ alongside a "US" transcription like "/kət/" is likely to come away with the misimpression that there is a divide between the US and UK in the phonetic quality of the "strut" vowel, or that the accents use different vowel phonemes in such words. (While there are differences in the quality of "strut", I don't believe they split cleanly along national lines.)--Urszag (talk) 01:45, 30 December 2023 (UTC)

@Urszag: Again, I don't feel qualified to judge. 0DF (talk) 02:09, 30 December 2023 (UTC)

There is a difference in the phonemic representation though. (And I would argue that there is a difference between GA & RP in terms of phonetic quality as well) @-sche: I really don't see why we'd need to keep both standards in line? It'd be like arguing that we should have vowel length marked in GA because it's marked in RP (which a few sources do!), even though it's very much not in fashion anymore. More and more sources have switched to using /ə/ for GA instead of /ʌ/, leaving us now in a minority, so I'd really like to focus on that issue specifically. The issue of RP vowels can be dealt with at another time. We should treat each dialect as if they're the only one we show. I know you already know this, but I'd also like to remind folks in general that GA representation is not intended to represent how every American speaks, as I feel that that's something that's been heavily conflated as of late. AG202 (talk) 02:33, 30 December 2023 (UTC)

I don't think that's a good argument because of how we represent long Os (as in doe, though and grow); different symbols are used for it in GenAm and in RP (oʊ vs əʊ) even though they're the same phoneme. 2804:1B0:1903:CDA7:A940:DFBD:D623:EFA6 06:31, 25 January 2024 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘While I don't agree with the first part, I agree with the second (not every American speaks GenAm). However I vote, it's been striking during all the discussions that much of the push for change has been from people citing a guy who measured how speakers produce the formants and whether they could reliably identify something as NURSE or STRUT vs schwa ... and much of the opposition has been people saying "no, the schwa in the first syllable and the STRUT in the second syllable of undone are separate vowels for me", or "I pronounce the a in Rosa's with /ʌ/ and I distinguish that from a schwa", or "I pronounce en- with a schwa, and it's different from un-".
But (as you know!) anyone whose accent actually distinguishes schwa-vs-/ʌ/ knows both vowels in undone are /ʌ/. And Rosa's is not /ʌ/. And anyone who speaks GenAm, or has looked at reference works about it, knows en- is not a schwa in GenAm. But our perceptions are colored by our expectations, and everyone assumes they speak with the standard accent ... and wants the symbols used to notate the standard accent to match their not-scientifically-measured perception of their own speech. I have flashbacks to the [different!] people who insisted that GenAm and "most Americans" phonemically merged pull, pool, pole and purple all to /l̩/, or merged cot and caught to /ɒ/. No, most Americans don't merge those sounds, and most of those who do don't merge them to that! Some speakers and/or local or regional (sub-)accents do, but those aren't GenAm. - -sche (discuss) 03:29, 30 December 2023 (UTC)

I mean at this rate it's not just Geoff Lindsey who's arguing for schwa but the OED & MW and other sources I've mentioned before. But yeah, in that case, I don't really get what the other opposition is then other than "it's what we've done till now", which, while very common on Wiktionary, shouldn't be the standard. It's led us to fall behind many times compared to other reputable sources. AG202 (talk) 03:34, 30 December 2023 (UTC)

@-sche: I'd like to request a removal of the starting date and ending date for the time being, at least as someone in support of the proposal. I'm a big proponent of having votes be initiated by the folks that actually want the proposal, even if I oppose it, as it makes sure that it has the proper background and facts before possibly shutting it out for a much longer time, such as with Wiktionary:Votes/pl-2023-02/THUB amendment for example. As of right now, I don't think this vote is in the right shape to do that. AG202 (talk) 05:50, 30 December 2023 (UTC)

I’d like to voice my opposition to your request, I hope the vote goes ahead and I shall be voting in favour when the vote does start. Clearly most people in Britain and America use the same quality of vowel for CUT as they do for NURSE, despite the length distinction, and many even use the same vowel for COMMA (though that is technically a separate issue). There are people who use a different vowel for NURSE in South Wales, Australia and New Zealand and sometimes even in North West England and the poshest of the poshest of old-fashioned RP that hasn’t been spoken in its purest form for decades (think Celia Johnson in Brief Encounter) but they’re hardly representative of Britain and certainly not of America. Overlordnat1 (talk) 07:28, 30 December 2023 (UTC)

I dont see anywhere in the linked discussion where undone is described as having two separate vowels. That would be inconsistent with distinguishing en- vs un-, for one thing. One reason Im so strongly against this proposal is that i think not only that it's factually wrong, but that the people proposing it dont properly understand the opposing side. —Soap— 05:01, 30 December 2023 (UTC)

As well, I see the people in support of this reform also disagree with each other, and that pushing it through will only result in more conflict. But oh well. —Soap— 10:20, 30 December 2023 (UTC)

I kind of think the entire concept and terminology of "General American" is outdated and perhaps was fake to begin with. I understand that it is problematic to abandon the concept, since there clearly are accents in the US that are regional and/or viewed as 'nonstandard' and we don't want either to erase those accents (by presenting an overly narrow, prescriptive transcription under the broad label of 'US') or to present a confusing and inconsistent mixture of the diverse accents spoken in the US. But I feel like we should attempt, as best as possible, to give transcriptions that guide readers to a relatively neutral contemporary US accent, rather than focusing on providing transcriptions of a specific accent that it is imagined was spoken in the Western and Midwestern US a century ago—which is what "General American" suggests to me, especially when pronouncements like "en- is not a schwa in GenAm" are bandied about as authoritative statements. (There are similar problems of "Received Pronunciation" as a term that was originally introduced to describe a contemporary accent, but that now is either outdated as a reference point for modern learners, or unclear in what it refers to if repurposed to apply to some present-day accent). The Oxford English Dictionary does not refrain from offering "/ᵻnˈd(j)ʊ(ə)r/" "uhn-DYOOR" and "/ənˈtaɪ(ə)r/" "uhn-TIGH-uhr" as possible "U.S. English" pronunciations of endure and entire. Not saying that the OED is necessarily making the right decision in using these representations, but I just think it shows that it isn't so outlandish to view the use of /ə/ in this context as falling within the bounds of normative present-day US English (to the extent that nationwide norms are even a coherent concept).--Urszag (talk) 07:52, 30 December 2023 (UTC)

Extra clarity on phonemic vs. phonetic representation

I assume the vote is about phonemes (something you'd put between /forward slashes/) rather than (allo)phones (what you'd put between [square brackets]). Whether my assumption is correct or not, it would be helpful to clarify this in the vote's text. Chernorizets (talk) 11:45, 30 December 2023 (UTC)

The text about the vote consistently uses slashes ("writing the STRUT vowel as /ə/ (instead of /ʌ/), and the NURSE vowel as /ɚ/ or /əɹ/ (instead of /ɜ(ː)/~/ɝ/)"). I would say it is about the notation used for phonemes as much as it is about phonemes themselves, but it seems clear to me that the vote as currently written does not apply to phonetic transcriptions.--Urszag (talk) 11:57, 30 December 2023 (UTC)

Separate proposals for the NURSE and STRUT vowels individually

Shouldn't we be voting on whether to represent the NURSE vowel as schwa, and on whether to represent the STRUT vowel as schwa, separately? It might well turn out that it's appropriate to represent one of the two as schwa, but not the other. Whoop whoop pull up ^{Bitching Betty ⚧️ Averted crashes} 21:33, 3 January 2024 (UTC)

As far as I have seen in all the discussions of this, reference works and authorities either take both to have merged, or continue to use the historical notation for both. So far, it has appeared that the only people who think only one or the other of the two but not both have merged are a couple of Wiktionarians with (AFAICT) no phonological training, who happen (in other cases) to have demonstrably wrong perceptions of how 'most Americans' speak (which is understandable, because it can notoriously be hard for speakers to hear slight differences, in the way of mergers or splits, that they don't themselves make—expectation colors perception—and to notice if their own accent differs from the norm. this is not to say anyone is speaking wrong! but when we're deciding how to represent how most Americans speak, whether we call that GenAm or some new label, it seems more useful to stick to references which have measured how most Americans speak, vs to the lay assessments of a few of us editors!). If anyone can bring forward references that say one of the two has stayed distinct while the other has merged, that would be most helpful. - -sche (discuss) 21:56, 3 January 2024 (UTC)

They are two different situations ... so far as I know, there never was a phonemic contrast between /ɜr/ and /ər/ in any dialect of English, whereas the STRUT case is based on the belief that a phonemic contrast between /ə/ and /ʌ/ has collapsed in AmEng and AmEng only. They are not the same situation at all, and one cannot follow from the other. This looks to me like a way to avoid the more difficult STRUT debate by focusing on the easier NURSE debate, which very few people would disagree with, and then saying that if we've proven one case we've proven the other. —Soap— 13:55, 4 January 2024 (UTC)

Actually, now that I think of it, I have to assume you're calling me out in the paragraph above as someone who, in your words,

[has] no phonological training, who happen[s] (in other cases) to have demonstrably wrong perceptions of how 'most Americans' speak

No, Im not a professional linguist. And I dont think that matters. We should write our policies based on the actual issues at hand, not on who is speaking from a position of higher authority. —Soap— 14:11, 4 January 2024 (UTC)

We have to work based on consensus regardless. If we just allowed everyone to have their own ideas of how they pronounce things or how they define things, then we could have an infinite amount of pronunciations or definitions based on how individuals perceive things. That's why we source and point to established literature on the topic rather than solely rely on our own anecdotal experiences. AG202 (talk) 20:44, 4 January 2024 (UTC)

I think it would be beneficial if the vote provides references to the literature. For example, Merriam-Webster renders STRUT as strət - I assume that's not an arbitrary choice, and based on something in the literature. I believe adding this convenience over having to spelunk in the linked discussions to find those reference sources would help field questions on authority. Chernorizets (talk) 02:24, 6 January 2024 (UTC)

I agree, it's part of why I said that the vote should be postponed to give enough time to gather those resources, @-sche. Personally, I frankly don't have enough time to devote to that effort before the vote is slated to begin. AG202 (talk) 03:26, 6 January 2024 (UTC)

I did remove the start date (which I figured postponed it indefinitely) back in December when you suggested that. I'll remove it from the watchlist-list now, although this makes it less noticeable. - -sche (discuss) 08:06, 6 January 2024 (UTC)

I'd certainly vote for the nurse proposal, but am unsure about the strut proposal. It seems obvious to me that before /ɹ/ there's just one schwa-ish phoneme in General American. (The case might be different in some Southern and Eastern American accents, which could use a different set of symbols, though it's a shame we don't have any tables laying out what those symbols should be for anything but General American.) Not before /ɹ/, I think that General American doesn't have three schwa-ish phonemes /ɪ ə ʌ/ with the same distribution as old-fashioned RP, and there are probably just two, one higher than the other. That means it's a choice between /ɪ ə/ and /ə ʌ/ if we use the traditional RP symbols to represent the somewhat different General American phonemes.

But I'm not sure whether stressed /ʌ/ should be identified as the same phoneme as unstressed schwa as in the proposal. In my own speech, which might match all General-American-ish speakers in this respect, a stressed schwa seems to match traditional stressed /ɪ/ words in one example: just (adverb) when stressed for emphasis matches gist and has a higher vowel than just (adjective), which matches other cases of stressed strut. I guess this is because I have a relatively low kit vowel that is basically mid-central (where ə is usually put on the vowel chart). So I like the idea of transcribing kit with /ə/ and strut with /ʌ/ (or /ɐ/). Nurse and letter don't feel as low as strut so I end up grouping them with schwa and kit. Final schwa like in comma might belong to either phoneme, but I don't think I have a contrast between a high and low schwa at the end of a word. But it seems like based on prior discussions, other people would feel that just (adverb) and many other cases of schwa should be written with /ɪ/, so I may be the only one with this distribution (though it is apparently very similar to the distribution in New Zealand English, which my accent is otherwise very different from).

By the way, the choice of /ə ʌ/ also makes it easy to represent the contrast of furry and hurry in some eastern American accents as /əɹ/ versus /ʌɹ/. But that's not strictly General American.

I said "not before /ɹ/" because while I'd write most traditional stressed /ɪ/ as /ə/, the near and mirror vowel is different and I'd write it as /iɹ/.

Splitting the poll makes sense to me because it's a much more clear-cut case before /ɹ/. Many American speakers contrast tense and lax vowels in most environments, but have a smaller set of vowels before /ɹ/, where for many highly merging speakers you can only identify five more peripheral vowels /a e i o u/ (not usually written with those exact symbols in Wiktionary) and a schwa. Not before /ɹ/ there are at least two schwa-ish phonemes and it's more controversial how to write them. I'd love it if we could agree on which of the schwas-not-before-/ɹ/ to identify with the single schwa-before-/ɹ/, but there seems to be more disagreement on that. It would at least be an improvement to no longer give the impression that people need to distinguish two schwa-ish phonemes in GA before /ɹ/. — Eru·tuon 19:29, 25 January 2024 (UTC)

As another GA speaker, I would not support transcribing "kit" with /ə/. That's a very different vowel to me. Stressed just (adverb) does not change to the /ɪ/ vowel found in gist. I've heard it before, but primarily from folks that don't speak General American. I also don't really see any literature that supports representing kit with a schwa for GA. It feels dialectal/idiolectal. AG202 (talk) 19:42, 25 January 2024 (UTC)

(I was going to avoid continuing to say this, but was prompted by something in the BP:) This is why I keep wanting to defer to reference works that actually measure what sounds people actually produce and can distinguish, because well-meaning people keep saying (in good faith!) "I and most Americans say X", where X is something most Americans do not say. Our brains are over-good at ignoring all kinds of differences in how people speak, because they have to ignore all kinds of differences in the frequency (etc) of e.g. one man's /u/ and another woman's /u/ that come from them simply having different voices, when letting us know that both people said /u/, and so people with (e.g.) the cot-caught-merger often also ignore the differences made by people who do distinguish those sounds, and vice versa (people often don't realize they're speaking differently, unless specifically listening for it, and sometimes not even then). What we hear is colored by what we expect to hear, we sort what we hear into the inventory of phonemes we expect or are used to, and it's human nature that most of us think we speak the standard or normal way... even when we don't. E.g., you say you speak GenAm, and then say you have a schwa in kit words and [in the BP] say you have or, in the BP, another user has an /ɒ/ in caught words, without noticing(?) that that is not GenAm and is (AFAICT) not something most Americans do. (Naturally, people opposed to representing the nurse vowel as schwa must think proponents are similarly failing to hear the distinction. But that's why I think it's useful to defer to data rather than my or anyone else's lay assessment.) - -sche (discuss) 08:25, 9 February 2024 (UTC) - -sche (discuss) 19:19, 11 February 2024 (UTC)!

@-sche Are you sure that [ɒ] is not something most Americans with a cot-caught distinction say for caught? I know the vowel normally used to represent this is /ɔ/ but that is just a convention. For example, the Wikipedia article on Open back rounded vowel specifically says this vowel occurs in General American in the thought series with the comment "Vowel /ɔ(:)/ is lowered (phonetic realization of /ɔ(:)/ is much lower in GA than in RP)". This is accompanied by an audio sample [1] that sounds exactly like the way I say thought. My [ɒ] in caught and thought has to my ears the same quality as RP's [ɒ] in cot and thot except that the RP vowel is shorter (i.e. mine is short and RP's is ultra-short, not that I have a long vowel here). This is IMO completely different from the issue of pronouncing kit with a /ə/, which is not remotely similar to any vowel I've ever heard Americans pronounce kit with. Benwing2 (talk) 11:00, 9 February 2024 (UTC)

I noticed that w:Inland Northern American English mentions lowering of kit, so maybe that feature of my accent is a regional thing (less noticeable than cot sounding like some other dialects' cat, which isn't a feature I have). But you've prompted me to check Wikipedia and actually File:Mid front unrounded vowel.ogg and File:Close-mid central unrounded vowel.ogg sound more like my possible kit pronunciations than File:Mid-central vowel.ogg. And I recalled that I have another sort of relevant definitely non-GA pronunciation: a parallel to Canadian raising in /ɑɹ/, at least before voiceless consonants in the coda (heart), distinct from the opener vowel in hard and the closer vowel in hurt, and more similar to strut than kit. (Maybe occasionally not determined by environment, like Canadian raising.) So I've clearly got a non-GA idiolect and I'm sometimes inaccurate at measuring my own vowels. This inclines me more to accept the proposal if it's what academic works support. — Eru·tuon 04:40, 28 February 2024 (UTC)

@Erutuon I have the same "Canadian raising in /ɑɹ/" in my speech as well, making Carter and carder non-homophonous, so I suspect it's less idiolectal than you think. Benwing2 (talk) 04:44, 28 February 2024 (UTC)

Miscellaneous references

Since some comments above have argued that we should appeal to authority in this topic, here are some quotes. What John Wells says about this subject at the start of the Longman Pronunciation Dictionary (2000):

"LPD distinguishes between the vowels ʌ and ə, although in AmE they can generally be regarded as allophones of the same phoneme, and for some speakers are more or less identical phonetically too. Thus where LPD writes above ə ˈbʌv some speakers pronounce ə ˈbəv. Similarly LPD distinguishes between ɝː and ᵊr, as in further 'fɝːðᵊr, although many speakers have a similar syllabic [r] in both syllables. All these qualities arguably represent the same phoneme ə, with or without a following r.

Where RP has ʌr followed by a vowel sound, most Americans use ɝː and that is what is shown in LPD entries: courage ˈkʌr ɪdʒ ║ ˈkɝː-. It should be noted, however, that there are other Americans who use ʌr, as in RP." (p. xv)

A paper by Péter Szigetvári arguing that strut = letter/comma in British English: Stressed schwa in English (2018). Szigetvári points out the importance of transcribing stress on syllables other than the primary-stressed syllables in transcriptions that use one symbol for both strut and schwa.--Urszag (talk) 11:09, 6 January 2024 (UTC)

A phonetic study that cites Szigetvári 2018 says "unstressed [ə] may be produced [ɨ] in certain contexts, such as in the -ed suffix (e.g., wretched, blurted) or plural morpheme following sibilants (roses, see Flemming and Johnson, 2007). This may be evidence that the upper F1 category boundary may be comparable between stressed and unstressed [ə], but the lower F1 boundary may extend lower (i.e., encompass a greater range of acceptability) for the unstressed vowel. An overall category size that is greater for unstressed [ə] than stressed [ə] is also supported by accounts that this unstressed vowel is more variable in production than other vowels (Magen, 1984). This greater range of acceptable F1 values for unstressed [ə] would predict that there could be a greater range of produced F1 values in the unstressed variant as well. In productions of “above” from the baseline phase (unaltered auditory feedback) in both experiments, F1 values produced during the stressed vowel were a subset of those produced during the unstressed vowel. Indeed, the unstressed [ə] vowel contained lower values of F1 than those produced in the stressed vowel (Fig. 6)." (page 716, "Effects of syllable stress in adaptation to altered auditory feedback in vowels", Bakst, Sarah and Niziolek, Caroline A., J Acoust Soc Am. 2021 Jan;149(1):708. doi: 10.1121/10.0003052. PMID: 33514177; PMCID: PMC7846293.)

A thought that occurs to me after this last study is that using ə and ᵻ (but no ʌ) could be a potential alternative convention, although not necessarily a very defensible one (since I think there's no possible argument that ᵻ is phonemic, rather than diaphonemic.) Geoff Lindsay, in the description of his Youtube video ""Schwa is never stressed" – FALSE" (2022), writes "(Note: many Americans feel they have their unstressed /ɪ/ rather than /ə/ in some of the words shown from dictionaries in the video. But that still doesn't mean a distinct /ʌ/.)"--Urszag (talk) 11:22, 6 January 2024 (UTC)

Was there ever /ɜɹ/ or /ʌ/ in General American?

The vote currently says "Ideally, also recording the historical pronunciation (as /ʌ/ and /ɝ/, respectively) with some appropriate qualifier like ". It helps clarify that we're not banning those symbols from all accents' transcriptions, but was /ɝ/ or /ɜɹ/ ever distinct from /əɹ/, or /ʌ/ from /ə/, in accents that eventually gave rise to the modern General American accents that don't have these distinctions? I'd rather not have the vote assume a conclusion either way.

I read A. Z. Foreman (who analyzes historical phonetic treatises) saying that Benjamin Franklin (who might have had a pronunciation close to the ancestor of General American) identified strut with schwa already in the 1700s. It's possible that when strut split from foot, some accents merged strut with schwa immediately and kept it merged, while others (such as whatever accents gave rise to old-fashioned Received Pronunciation) maintained some sort of distinction.

I would be fine with the statement that we're only voting on symbols for modern General American phonemes and the symbols /ʌ/ and /ɝ/ (or /ɜɹ/) can be used to transcribe other accents' phonemes where appropriate. — Eru·tuon 20:10, 25 January 2024 (UTC)

I also am skeptical of the accuracy of describing /ʌ/ and /ɝ/ as specifically "historical" or "older" pronunciations (and on the other side of the coin, the idea that "/ə/" is more "current" or "up-to-date"). I posted this already in the December 2023 Beer parlour, but Daniel Jones's transcription in the Phonetic Dictionary of the English Language (1913) marked the vowel in "earl" as əː (although Jones also used u, i, ɔ for what Gimson would transcribe as ʊ, ɪ, ɔ, which implies some room for slight quality differences).--Urszag (talk) 02:07, 30 December 2023 (UTC)

Also, John Walker was identifying the vowel of tub with the unstressed vowels in the last syllables of disaccustom and director in 1791. I realize these two examples aren't "General American" but my point is that the fact that different notations have gone in and out of fashion doesn't mean there was necessarily a corresponding change in either the phones or phonemes that people were actually using in the pronunciation of words like this.--Urszag (talk) 21:05, 25 January 2024 (UTC)