English Dialects

[abi]

Strange to see so many americans here with [ɚ] over a hard [ɻ] as I've always considered myself to have a bland, typical "midwestern" accent despite never having actually lived in the midwest (the few times I've been to Michigan people called me out on vocab like "y'all" and "soda" rather than accent). I remember having such a hard time learning to roll my r's because I kept trying to vibrate my tongue with it anchored to the roof of my mouth! I'm also 90% sure my /ɘ/ is always [ʌ] or [ɪ] and never truly central: luxury [lʌk.'ʃɻ.i], delicious [dʌ.'lɪʃ.ɪs]

KIT ɪ
DRESS ɛ
TRAP æ
LOT ɑ
STRUT ʌ
FOOT ʊ
BATH æ
CLOTH ɑ
NURSE ɻ
FLEECE i
FACE ɛi
PALM ɑ
THOUGHT ɑ
GOAT ɔɯ
GOOSE ɯ
PRICE ɑi
CHOICE ɔi
MOUTH ɑɯ
NEAR iɻ
SQUARE ɛɻ
START ɑɻ
NORTH ɔɻ
FORCE ɔɻ
CURE jɻ
happY i
lettER ɻ
commA ʌ

[Xonen]

Strange to see so many americans here with [ɚ] over a hard [ɻ]

I suspect this is a question of notation more than pronunciation; people just prefer to use the symbol /ɚ/ for the syllabic version (since that seems to be pretty much the standard for American accents) rather than try to figure out exactly what each of the different articulators is doing.

[Khemehekis]

What, as in the TRAP-BATH split?

(There are some old dictionaries where the pronunciation guides seem to indicate the existence of the TRAP-BATH split in AmE, with slightly different conditioning than the British split (it applied to 'ant'), but it died out and for all I know was artificial in the first place.)

(Could the variance in the pronunciation of 'aunt' be a relic of the split? It can't be a spelling pronunciation; otherwise we'd expect THOUGHT instead of PALM.)

Yes, the trap-bath split.

I have [æ] for "ant", [a] for "aunt", [æ] for "trap", [a] for "bath", [ɑ] for "thought", and [ɑ] for "palm". Make of that what you will.

By artificial, do you mean affected?

(I know a woman who says [dɑns] when imitating the way I pronounce "dance" ([dans]), and says her son "hates pronunciations like that".)

[Nortaneous]

I have [æ] for "ant", [a] for "aunt", [æ] for "trap", [a] for "bath", [ɑ] for "thought", and [ɑ] for "palm". Make of that what you will.

Does "palm" have /l/? Which vowel do you have in "father"? How common is this?

By artificial, do you mean affected?

Yes, imported from Britain.

[Khemehekis]

I have [æ] for "ant", [a] for "aunt", [æ] for "trap", [a] for "bath", [ɑ] for "thought", and [ɑ] for "palm". Make of that what you will.

Does "palm" have /l/?

Yes. I have a /l/ in "palm", "calm" and "qualm", but not in "walk", "talk", "chalk", "stalk", "folk", "yolk", "half" nor "calf".

Which vowel do you have in "father"?

I have [ɑ] in "father". Also in "bother".

How common is this?

Not many people around me speak like this. I picked up some time around 12-14. At age 14, when I went to the National Spelling Bee, I remember Ned G. Andrews pronouncing the winning word [ˌanthədəˈluviən] in his Tennessee accent. And today, I have a Paratransit driver who wishes me a good [ˌaftɚˈnun]. To people who don't speak like this, it sounds like either [æ] or [ɑ] to them, judging by the way they imitate my speech.

[Nortaneous]

So... what all are the possible low vowel outcomes in the English dialects?

Code: Select all

       TRAP BATH MAD FATHER LOT CLOTH THOUGHT START
GenAm    æ    æ   æ     ɑ    ɑ    ɔ      ɔ      ar
PA/NJ    æ    æ   eə    ɑ    ɑ    ɔ      ɔ      ar
Boston   æ    æ   æ     ɑ    ɔ?   ɔ?     ɔ      ɑ
WestAm   æ    æ   æ     ɑ    ɑ    ɑ      ɑ      ar
RP       æ    ɑ   æ     ɑ    ɒ    ɒ      ɔ      ɑ  (+NORTH/FORCE = THOUGHT)
Kheme    æ    a   æ     ɑ    ɑ?   ɔ?     ɔ      ar? 
Newfndl  æ    æ   æ    æ/ɑ?  ɒ    ɑ      ɑ      ær	

[Reyzadren]

^If you want more samples for that table, here's mine: (Note: I identify as using BrE/RP, but this is not likely a pure version)

TRAP e
BATH a
MAD e
FATHER a
LOT ɔ
CLOTH ɔ
THOUGHT ɔ
START a

[Nortaneous]

LOT = THOUGHT?

[alynnidalar]

I speak Inland North American (with the Northern Cities Vowel Shift), so for me it's:

TRAP/BATH/MAD: eə
FATHER/LOT: a (well, ä, I think?)
CLOTH/THOUGHT: ɒ
START: ɐɹ

[Salmoneus]

So... what all are the possible low vowel outcomes in the English dialects?

In any dialect anywhere? There's a lot of possibilities...

It may be easier to understand seen as two different processes - the cloth instability, and the trap instability. But first, /a/ is lengthened a) in 'father', b) before coda /r/, and c) compensating for the loss of /l/ before consonants (PALM, etc).


Cloth instability

1. The lot/cloth split creates two different phonemes, distinct from both THOUGHT and PALM.

2. In the US, but not in the UK, the cloth split is extended to pre-velar contexts in closed syllables, but not all of them.

3. A small number of additional words show cloth-lengthening in some of the US, but not everywhere. Most famous is the CLOTHing of "on" in midland and southern US dialects.

4. Many dialects, including RP and US dialects, merge CLOTH with THOUGHT, bringing the phoneme count down to three. Crucially this does not happen in the ancestor of SSBE.

5. Many dialects further increase the distinction between LOT and CLOTH by derounding the former. This does NOT happen in RP or SSBE, but does happen in the West Country, several other rural English dialects, in Ireland, and across most of the US, but NOT in Boston. [sociologically, we might see this as general derounding of LOT, with conservative resistence in high-status urban dialects on both continents].

6. This derounding brings LOT toward PALM. The distinction is retained in the UK, but in the US the distinction is lost and there's a LOT/PALM merger. This gives the GA two-phoneme outcome, PALM/LOT vs CLOTH/THOUGHT.

7. Many dialects finally surrender the LOT/CLOTH split. The two big exceptions to this are RP and GA. However, because of the previous couple of mergers, the outcome of LOT-CLOTH merger is different in different areas. In Scotland and in Boston, you get PALM vs LOT/CLOTH/THOUGHT. In SSBE, where we didn't have the CLOTH-THOUGHT merger, you get a three-phoneme outcome, PALM vs LOT/CLOTH vs THOUGHT. In cot/caught-merging parts of the US, you get a one-phoneme outcome, PALM/LOT/CLOTH/THOUGHT.

8. Most dialects outside of very rural areas, FORCE then merges with the vowel of THOUGHT/NORTH.

9. In London, THOUGHT then splits (potentially kicking off a new cycle?).


Trap instability

1. As discussed, /a/ lengthens in START, PALM and FATHER.

2. In southern British, and descendent colonial dialects (US, australia, etc), TRAP is further tensed before coda voiceless fricatives, and inchoately before coda nasals in clusters, creating BATH.

3. BATH expands to include all coda nasals in clusters, except in Australia.

4. In the southern UK outside of the west, in Boston, and in the southern hemisphere, BATH merges with PALM. In the West Country, and in the US outside of Boston, BATH remains distinct. However, in the US in particular there can be some irregularity - some BATH words can merger with PALM (most commonly AUNT), while some PALM words can instead join BATH (most commonly HALF, CALF).

5. However, tensing/splitting continues. Surviving BATH merges into a new set we might call CAN. Tensing continues to before all coda nasals - except that function words with simple codas are immune. In the West Country, BATH remains distinct from both CAN and PALM. [producing, along with PALM and derounded LOT, no fewer than FOUR unrounded low vowels!].

6. Tensing spreads to the words 'mad', 'glad' and 'bad', and sometimes 'sad', and their derivatives.

7. Tensing spreads to before coda /g/ in Canada and parts of northeast US (mirroring the spread of cloth-lengthening to pre-velar contexts in these dialects).

8. In the US, outside of a NYC-Philadelphia-Baltimore region, and outside Cincinnati and New Orleans, tensing spreads to all pre-nasal vowels, even in open syllables.

9. In some areas of the US, tensing spreads to function words

10. In the inland north area, tensing spreads to all words

11. In the UK, at some point in the 20th century, CAN is lost, becoming a subphonemic variation tendency. It remains salient in Australia, mostly as a length distinction.

12. In the US, CAN is raised, and in most areas is broken into /E@/, /e@/ or /I@/, sometimes with additing raising of the second element to /eI/ in some contexts.

13. In some areas of the US that still have a CAN/TRAP split, CAN spreads to include more words before voiced stop codas.



Finally, in some parts of the (western) US and Canada, where the cot/caught merger has resulted in a rather large gap, TRAP/CAN slides backward.



....I'm sure there are some errors here, but it seems to be the general pattern?

[Ælfwine]

Mine:

TRAP/BATH/MAD: æ ("MAD" can be slightly diphthongized)
FATHER/LOT: ɑ
CLOTH/THOUGHT: ɑ
START: ɐɹ

[Frislander]

Sal you've missed the Geordyism of START/PALM/FATHER being rounded and near-to-totally-merging with THOUGHT.

[Nortaneous]

In any dialect anywhere? There's a lot of possibilities...

Not all dialects need to be descended from the same maximally recent common ancestor. Is there a reason to posit a 'surrender' of the LOT-CLOTH split, rather than saying it just didn't happen in those dialects in the first place? And why should the TRAP-BATH split be related to the TRAP-MAD split? /{/ is an unstable vowel anyway (cf. Slavic yat), and the English low vowel system is clearly unstable.

Then again, American English probably did have the TRAP-BATH split (with BATH sometimes merging into LOT/FATHER and sometimes not) and mostly reverse it later -- it shows up in some old Webster dictionaries and Deseret texts, at least.

[Salmoneus]

In any dialect anywhere? There's a lot of possibilities...

Not all dialects need to be descended from the same maximally recent common ancestor.

But where dialects can be described as being descended from a common ancestor through simple sound changes, it's parsimonious to do so.

Is there a reason to posit a 'surrender' of the LOT-CLOTH split, rather than saying it just didn't happen in those dialects in the first place?

Parsimony. This enables us to posit a single common change throughout southern English before the dialects had fully diverged, and then a single change later on spread by areal influence.

The alternative is to a) posit the change happening or not, randomly, in otherwise similar and related dialects, and then all the dialects independently undergoing parallel sound changes for no apparent reason that exactly mimic the effect of merging LOT and CLOTH and, crucially, whatever sets those two have already merged into.

Take cot-caught merged American. We have two possible models:

A:
- like other southern englishes, all american splits LOT and CLOTH
- like many other englishes, including RP, CLOTH merges with THOUGHT
- like other west country englishes (the english dialects american outside of boston is most similar to), LOT derounds
- derounded LOT merges with PALM
- as in Scottish English, LOT/PALM merges with CLOTH/THOUGHT (happening particularly in scottish-influenced areas)
- this final merger can gradually spread through other american dialects

or
B:
- way back in the 17th century, American splits into the dialect that will end up cot/caught merged and the one that won't
- in the merging dialect, LOT and CLOTH don't split
- LOT/CLOTH derounds and merges with PALM
- LOT/CLOTH/PALM for some reason merges with THOUGHT, a novel change for no apparent reason and not mirrored elsewhere
- the spread of the cot/caught merger is then actually the replacement of original lot/cloth-splitting dialects with almost-identical dialects that actually instead have remained separate for centuries and just happen to look the same, except for the PALM/THOUGHT merger.

B seems less simple than A.

Similarly, the hypothesis of a general, cross-dialectical collapse of the distinction between LOT and CLOTH can explain why cot/caught in the US is paralleled by the otherwise unrelated rise of SSBE in the UK. We can assume two parallel dialects in the south for four hundred years - one splitting, one not splitting - with the non-splitting dialect just coincidentally happening to suddenly replace the otherwise almost identical splitting dialect in the UK at the exact same time that a non-splitting dialect is gaining ground rapidly in the US. Or, we can just imagine that the split is lost at a similar time on both sides of the Atlantic.

And why should the TRAP-BATH split be related to the TRAP-MAD split?

Parsimony. Sure, we could assume that there are half a dozen different a-tensing sound changes occuring continually but unrelatedly across many continents that just happen to share the same conditions. Or, we could assume that there's a single ongoing process of a-tensing across a spreading set of conditions, which happens in some dialects to hav been interupted by backing of the tensed form.

So, to take the American case again, we have two models:

- the BATH vowels are tensed across southern English
- the BATH vowels are backed in RP, but not in western/american
- BATH-backing spreads to Boston, and begins to spread across America, but only reaches a small number of words
- BATH-tensing continues through a series of new conditions, leaving behind a number of conservative areas that don't expand tensing to all these conditions

OR:

- the BATH vowels are tensed and backed in RP
- the BATH vowels are independently tensed but not backed in the west
- totally unrelatedly, new sound changes happen independently in New York, the mid-atlantic region, cincinnati, new orleans, the great lakes region, and independently in "all the places that aren't thos places" that all just happen to involve tensing (but not backing) of the BATH set
- simultaneously and unrelatedly, parallel sound changes that also involve tensing of the same vowel in the same condition happen in australia and (subphonemically, or temporarily) in southern england.
- the BATH set is independently tensed and backed in Boston
- a couple of words in the BATH set are randomly, independently backed across America

Personally, I think the "the BATH set is tensed all along, but just isn't fully backed in most of the US" is much, much simpler as an explanation.


Plus, we then get to theorise two parallel changes in very similar conditions to both low short vowels (LOT and TRAP) happening across southern english at the same time.

Then again, American English probably did have the TRAP-BATH split (with BATH sometimes merging into LOT/FATHER and sometimes not) and mostly reverse it later -- it shows up in some old Webster dictionaries and Deseret texts, at least.

I didn't know this, but it doesn't surprise me. Of course, you could still theorise that the BATH vowels are tensed, then untensed, then tensed again. But again, this doesn't seem the simplest solution...

[Khemehekis]

So... what all are the possible low vowel outcomes in the English dialects?

Code: Select all

       TRAP BATH MAD FATHER LOT CLOTH THOUGHT START
GenAm    æ    æ   æ     ɑ    ɑ    ɔ      ɔ      ar
PA/NJ    æ    æ   eə    ɑ    ɑ    ɔ      ɔ      ar
Boston   æ    æ   æ     ɑ    ɔ?   ɔ?     ɔ      ɑ
WestAm   æ    æ   æ     ɑ    ɑ    ɑ      ɑ      ar
RP       æ    ɑ   æ     ɑ    ɒ    ɒ      ɔ      ɑ  (+NORTH/FORCE = THOUGHT)
Kheme    æ    a   æ     ɑ    ɑ?   ɔ?     ɔ      ar? 
Newfndl  æ    æ   æ    æ/ɑ?  ɒ    ɑ      ɑ      ær	

My vowels are here:

viewtopic.php?f=8&t=6566&p=281983&hilit ... B9#p281983

[All4Ɇn]

KIT ɪ
DRESS ɛ
TRAP æ
LOT ɑ
STRUT ɜ
FOOT ʊ
BATH æ
CLOTH ɑ (not really sure very similar to palm but still different)
NURSE ɚ
FLEECE i/ɪi
FACE eɪ
PALM ɑ
THOUGHT ɑ (same as cloth)
GOAT ɜʊ
GOOSE ʊu
PRICE äɪ/äː
CHOICE ɔɪ
MOUTH æu
NEAR iɚ
SQUARE eɚ
START ɑɚ
NORTH ɔɚ
FORCE ɔɚ
CURE jɚ/jɔɚ
happY i
lettER ɚ
CommA ə

[Jampot911]

Mine's a Scottish accent, though one with a slight Doric hint (since I'm from the North-East), though more general since my parents come from different parts of Scotland.

KIT ɪ
DRESS ɛ
TRAP a
LOT ɔ
STRUT ʌ
FOOT ʏ
BATH a
CLOTH ɔ
NURSE ɪɚ
FLEECE i
FACE e
PALM a
THOUGHT ɔ
GOAT o
GOOSE ʏ
PRICE ʌɪ (not quite sure, but definitely not my standard 'a')
CHOICE ɔɪ
MOUTH ʌʉ (see PRICE)
NEAR iɚ
SQUARE eɚ
START ʌɚ
NORTH ɔɚ
FORCE oɚ
CURE jʉɚ
happY e
lettER ɚ
CommA ə

[Nortaneous]

Take cot-caught merged American. We have two possible models:

A:
- like other southern englishes, all american splits LOT and CLOTH
- like many other englishes, including RP, CLOTH merges with THOUGHT
- like other west country englishes (the english dialects american outside of boston is most similar to), LOT derounds
- derounded LOT merges with PALM
- as in Scottish English, LOT/PALM merges with CLOTH/THOUGHT (happening particularly in scottish-influenced areas)
- this final merger can gradually spread through other american dialects

or
B:
- way back in the 17th century, American splits into the dialect that will end up cot/caught merged and the one that won't
- in the merging dialect, LOT and CLOTH don't split
- LOT/CLOTH derounds and merges with PALM
- LOT/CLOTH/PALM for some reason merges with THOUGHT, a novel change for no apparent reason and not mirrored elsewhere
- the spread of the cot/caught merger is then actually the replacement of original lot/cloth-splitting dialects with almost-identical dialects that actually instead have remained separate for centuries and just happen to look the same, except for the PALM/THOUGHT merger.

B seems less simple than A.

The LOT-CLOTH split of course happened in cot-caught-merged AmEng. The sound change wasn't reversed. But for SSBE to have surrendered the split while still preserving THOUGHT as distinct from LOT and FATHER ( = PALM, but l-restoration is common nowadays), wouldn't it have had to have preserved unmerged CLOTH for hundreds of years? Unless there's evidence that it did, it seems better to say that the split just never reached fixation, or present-day SSBE is descended from (or at least influenced by?) a dialect that never underwent the split.

Then again, American English probably did have the TRAP-BATH split (with BATH sometimes merging into LOT/FATHER and sometimes not) and mostly reverse it later -- it shows up in some old Webster dictionaries and Deseret texts, at least.

I didn't know this, but it doesn't surprise me. Of course, you could still theorise that the BATH vowels are tensed, then untensed, then tensed again. But again, this doesn't seem the simplest solution...

Are there any words that take irregular lengthening in both the Mid-Atlantic dialects with a phonemic TRAP-MAD split and SSBE?

In that old Webster dictionary (which distinguishes all of TRAP, FATHER, and BATH), "halve" is given with... FATHER. Weird!

[Salmoneus]

But for SSBE to have surrendered the split while still preserving THOUGHT as distinct from LOT and FATHER ( = PALM, but l-restoration is common nowadays), wouldn't it have had to have preserved unmerged CLOTH for hundreds of years? Unless there's evidence that it did, it seems better to say that the split just never reached fixation, or present-day SSBE is descended from (or at least influenced by?) a dialect that never underwent the split.

Well yes, that is a problem. It's possible that SSBE is indeed descended from a dialect that avoided the split. I think what's actually more likely to happen is that the split was completed, but was subsequently reversed, not phonologically, but lexically, due to influence from other English dialects that did not undergo the split. But I don't know if that's exactly what happened; you'd need to study historical dialectology more closely.

I'm not sure, however, what meaningful distinction is really being drawn between "surrendering" a split and a split "not reaching fixation".

Of course, the reality was probably more complex and multi-factorial than any simplistic model.

[NB, l-restoration is very uncommon here, outside of a few unusual words like 'falcon'.]

Are there any words that take irregular lengthening in both the Mid-Atlantic dialects with a phonemic TRAP-MAD split and SSBE?

In theory a good question - although of course it may just be that whatever feature caused an irregularity in one place was missing in another, or that some irregularities have been repaired through analogy in some dialects. That said, for the sake of interest...

The SSBE BATH split doesn't have many irregular BATHs, though it has a bunch of irregular TRAPs. Most of these, however, are loanwords, abbreviations or other more recent developments.

Some words that might be worth testing if you have a splitter to hand might be:
- gaff ['gaffe', on the other hand, is a later loan, which might be mixing things up]
- hath [however, this may be evidence of the 'not function words' already operating at the early stage of the split]
- words ending -astic
- words beginning asp-
- gasket
- bastion
- jasper
- crass, mass, alas,

However, I think that a lot of these DID join BATH, but have irregularly been reassigned to TRAP, presumably under the influence of other dialects, or because they look like or are loanwords (which almost always take TRAP in SSBE). This process is ongoing, and many words traditionally BATH in RP are now mostly or entirely TRAP in SSBE (blasphemy, Mass, plastic/elastic, masturbate, paschal, carafe); I suspect this change simply happened earlier for some other words.

Contrariwise, US dialects have a clear tendency to progressively generalise tensing, so it would be no surprise if any original irregular TRAPS had already been tensed even in dialects that haven't spread tensing to all /{/.

In the other direction, "rather" has BATH irregularly, but that's clearly just an analogy to FATHER and may not be old ('lather' can be either).

There's also "castle" and "fasten", which irregularly have BATH; however, the irregularity here is the later loss of conditioning /t/.

In that old Webster dictionary (which distinguishes all of TRAP, FATHER, and BATH), "halve" is given with... FATHER. Weird!

Not at all. "Halve" is a PALM word, which Webster would have merged with FATHER. The vocalisation of the /l/ led to a diphthong, later assimilated into a long vowel.

[spanick]

KIT ɪ
DRESS ɛ
TRAP æ
LOT ɑ
STRUT ʌ but phonetically probably closer to ɜ
FOOT ʊ
BATH æ
CLOTH ɑ
NURSE ɚ
FLEECE i
FACE eɪ
PALM ɑ
THOUGHT ɑ
GOAT ɔʊ
GOOSE ʉ
PRICE aɪ
CHOICE ɔɪ
MOUTH aʊ
NEAR iɚ
SQUARE eɚ
START ɑɚ
NORTH ɔɚ
FORCE ɔɚ
CURE jɚ
happY i
lettER ɚ
CommA ə

Some things not fully captured by this word list. I have the caught/cot merger as well as the pen/pin. ɪ and æ are raised before the velar nasal.

Here's a vowel chart of my vowels I made about 6 years ago in grad school.

Spoiler: