English questions

[Travis B.]

I found this on Mastodon:

https://universeodon.com/@LadyDragonfly ... 9255529296

Midwest Guide to Yes and No

Yeah = yes
No = no
Yeah, no = no
No yeah, yeah = yes
Yeah, no yeah! = very yes
No, yeah no = very no
Welp = yes

My question is, how accurate is that?

That is pretty accurate. Note that at least here in southeastern WI there are also the affirmative /jɑ/ [ja(ː)] and the negative /nɑ/ [na(ː)] and /næ/ [nɛ(ː)].

[Raphael]

That is pretty accurate. Note that at least here in southeastern WI there are also the affirmative /jɑ/ [ja(ː)] and the negative /nɑ/ [na(ː)] and /næ/ [nɛ(ː)].

Thank you!

[salem]

I found this on Mastodon:

https://universeodon.com/@LadyDragonfly ... 9255529296

Midwest Guide to Yes and No

Yeah = yes
No = no
Yeah, no = no
No yeah, yeah = yes
Yeah, no yeah! = very yes
No, yeah no = very no
Welp = yes

My question is, how accurate is that?

That is pretty accurate. Note that at least here in southeastern WI there are also the affirmative /jɑ/ [ja(ː)] and the negative /nɑ/ [na(ː)] and /næ/ [nɛ(ː)].

I should also add (having grown up in Milwaukee) that it's reasonably common to replace a second "yeah" with "for sure", ie "no yeah for sure" and "yeah no for sure", and so are "yeah for sure" and "no for sure" (both shades of "yes") on their own.

[alynnidalar]

Agreed, these are quite normal in Michigan as well. (with the exception of "welp = yes". I do say "welp" but it's more of an expression of giving up or "I dunno", I think)

[Raphael]

Before I saw the initial Mastodon post, I would have thought that "welp" means something like "yikes".

[Travis B.]

Agreed, these are quite normal in Michigan as well. (with the exception of "welp = yes". I do say "welp" but it's more of an expression of giving up or "I dunno", I think)

Welp to me is acknowledging the truth of something with a level of resignation, or if not a response to something just expressing resignation overall.

[Imralu]

Dog has CLOTH/THOUGHT in the NAE I am familiar with; I myself am used to [ɒ] for it. As mentioned, LOT/FATHER normally has [a] (but it has [ɑ] adjacent to /r w h kw gw/), PALM has [ɑ], and START has [ɑʁˤ].

Hmm, I don't mean the exact realisations. I just mean how to know which lexical group which one belongs to. I guess it's because you have the LOT-CLOTH split and then a CLOTH-THOUGHT merger and the PALM-LOT merger, so it just seems pretty random to me without any clues from the orthography because I don't have the LOT-CLOTH split. For me, LOT-CLOTH is generally just represented <o> in a closed syllable (or ... orthographically closed by a double letter) and THOUGHT-NORTH-FORCE is <au>, <aw> or <or> (but <or> or <orr> before a vowel in the same morpheme is usually the LOT-CLOTH vowel, as in forest, lorikeet, horror, oral). There are some exceptions, like "fault", "auction", "caustic" and "Aus", "Aussie", which have the LOT-CLOTH vowel (not "Australia" though - that's a schwa!) although my mum says a lot of those with the THOUGHT-NORTH-FORCE vowel (although she also says "stereo" with the NEAR vowel instead of the DRESS vowel, which is weird), and <a> basically becomes like <o> after <w> or before <l>, except that it doesn't really ever become the GOAT vowel, so "water" has the THOUGHT-NORTH-FORCE vowel, as does "wart". My issue is basically that which lexical set these words belong to in US dialects without the COT-CAUGHT merger is impossible for me to predict from spelling, so I just find it easier to assume that LOT-CLOTH and THOUGHT words are all just PALM words as though everyone has the full cot-caught and father-bother mergers there.

[Travis B.]

Dog has CLOTH/THOUGHT in the NAE I am familiar with; I myself am used to [ɒ] for it. As mentioned, LOT/FATHER normally has [a] (but it has [ɑ] adjacent to /r w h kw gw/), PALM has [ɑ], and START has [ɑʁˤ].

Hmm, I don't mean the exact realisations. I just mean how to know which lexical group which one belongs to. I guess it's because you have the LOT-CLOTH split and then a CLOTH-THOUGHT merger and the PALM-LOT merger, so it just seems pretty random to me without any clues from the orthography because I don't have the LOT-CLOTH split.

Normally CLOTH is found before voiceless fricatives and in gone, and NAE it is found before many velars, particularly voiced ones. In some NAE varieties it is also found in on. Thing is, it is not particularly consistent.

For me, LOT-CLOTH is generally just represented <o> in a closed syllable (or ... orthographically closed by a double letter) and THOUGHT-NORTH-FORCE is <au>, <aw> or <or> (but <or> or <orr> before a vowel in the same morpheme is usually the LOT-CLOTH vowel, as in forest, lorikeet, horror, oral).

All those <or> words are NORTH/FORCE to me.

There are some exceptions, like "fault", "auction", "caustic" and "Aus", "Aussie", which have the LOT-CLOTH vowel (not "Australia" though - that's a schwa!) although my mum says a lot of those with the THOUGHT-NORTH-FORCE vowel

Those are all THOUGHT/CLOTH words here.

(although she also says "stereo" with the NEAR vowel instead of the DRESS vowel, which is weird), and <a> basically becomes like <o> after <w> or before <l>, except that it doesn't really ever become the GOAT vowel, so "water" has the THOUGHT-NORTH-FORCE vowel, as does "wart".

Water has THOUGHT/CLOTH here while wart has NORTH/FORCE.

My issue is basically that which lexical set these words belong to in US dialects without the COT-CAUGHT merger is impossible for me to predict from spelling, so I just find it easier to assume that LOT-CLOTH and THOUGHT words are all just PALM words as though everyone has the full cot-caught and father-bother mergers there.

I know it's easier, but I really would prefer it personally if non-Americans did not assume or act as if standard American English were cot-caught merged. Dictionaries of "American English" which have said merger are a personal pet peeve of mine, as of course you can trivially derive said merger from the lack thereof but not vice versa.

[KathTheDragon]

Agreed, these are quite normal in Michigan as well. (with the exception of "welp = yes". I do say "welp" but it's more of an expression of giving up or "I dunno", I think)

This is broadly the case for me in the UK, too

[Imralu]

For me, LOT-CLOTH is generally just represented <o> in a closed syllable (or ... orthographically closed by a double letter) and THOUGHT-NORTH-FORCE is <au>, <aw> or <or> (but <or> or <orr> before a vowel in the same morpheme is usually the LOT-CLOTH vowel, as in forest, lorikeet, horror, oral).

All those <or> words are NORTH/FORCE to me.

You don't distinguish "oral" and "aural" do you? For me, they're with the LOT/CLOTH vowel and the THOUGHT/NORTH/FORCE vowel respectively. Or do you distinguish them a different way?

I know it's easier, but I really would prefer it personally if non-Americans did not assume or act as if standard American English were cot-caught merged.

I mean, I'd personally prefer it if non-Australians did not assume or act as if Australian English lacked the bad-lad split - Hollywood actors doing Australian accents is excruciating - and I'd also really love it if people from the US stopped insisting that [ʉ̯] is /r/, but you get what you're given. The fact that a lot of people do merge PALM-LOT-THOUGHT in the US is very easy for us to imitate and it's not a mischaracterisation - it only is if someone says "THIS IS HOW ALL AMERICANS SPEAK". Non-merged accents are much more difficult because we'd have to learn each word on a case-by-case basis because they don't correspond to the same lexical sets in our dialects (the same reason why hollywood actors never get the bad-lad thing right). Yeah, I'm with you on the dictionaries for that reason. Why give less information when you could give more? I like how IPA of English words is usually shown on Wikipedia, with as many cross-dialectic distinctions shown as possible (lacking marginal things like bad-lad and assuming a universal fern-fir-fur merger).

[Estav]

I know it's easier, but I really would prefer it personally if non-Americans did not assume or act as if standard American English were cot-caught merged. Dictionaries of "American English" which have said merger are a personal pet peeve of mine, as of course you can trivially derive said merger from the lack thereof but not vice versa.

Given that you described this as a pet peeve, I don't know how seriously I should take it, but what does "not assume or act as if" even mean in this context? That it annoys you if someone who isn't a native speaker of American English chooses to adopt or imitate an accent with the merger?

In my opinion, it's hardly more useful to state that "standard American English lacks the cot-caught merger"; or at least, it seems to me that is only true if you use a definition of "standard American English" that refers to a relatively irrelevant theoretical construct (the archetypal, perhaps mythical "General American" of about a century ago). My experience has been that the presence or absence of the merger is largely ignored—at least by those who have it, and we're hardly a negligible minority. I have a hard time imagining a situation where a speaker who has the merger by nature would feel the need to consciously adopt a pronunciation without it when speaking on television/radio/etc.

There's an argument to be made that "standard American English" is just a fairly incoherent and unhelpful concept, but if we choose to speak about it as a thing, I'd consider it most accurate to say that it includes varieties without and with the cot-caught merger. Of course it can be helpful to give unmerged pronunciations in dictionaries, but it isn't all upside, since there is also the possibility of causing confusion for speakers to whom it isn't relevant (this is a stronger consideration for distinctions like horse vs. hoarse or wine vs. whine that are now much less widespread, but could also be argued to belong or have belonged to "General American" or "standard American English").

I'd also really love it if people from the US stopped insisting that [ʉ̯] is /r/

What does this refer to?

[Travis B.]

For me, LOT-CLOTH is generally just represented <o> in a closed syllable (or ... orthographically closed by a double letter) and THOUGHT-NORTH-FORCE is <au>, <aw> or <or> (but <or> or <orr> before a vowel in the same morpheme is usually the LOT-CLOTH vowel, as in forest, lorikeet, horror, oral).

All those <or> words are NORTH/FORCE to me.

You don't distinguish "oral" and "aural" do you? For me, they're with the LOT/CLOTH vowel and the THOUGHT/NORTH/FORCE vowel respectively. Or do you distinguish them a different way?

Nope, they're both NORTH/FORCE - only a small set of words with ⟨orV⟩ in them are LOT in most NAE varieties, such as tomorrow, sorrow, and for many, sorry (but I personally have NORTH/FORCE in sorry). The exceptions are Canadian English varieties, which tend to have NORTH/FORCE in these words across the board, and some East Coast varieties, particularly amongst older speakers, where words such as Florida also have LOT.

I know it's easier, but I really would prefer it personally if non-Americans did not assume or act as if standard American English were cot-caught merged.

I mean, I'd personally prefer it if non-Australians did not assume or act as if Australian English lacked the bad-lad split - Hollywood actors doing Australian accents is excruciating - and I'd also really love it if people from the US stopped insisting that [ʉ̯] is /r/, but you get what you're given. The fact that a lot of people do merge PALM-LOT-THOUGHT in the US is very easy for us to imitate and it's not a mischaracterisation - it only is if someone says "THIS IS HOW ALL AMERICANS SPEAK". Non-merged accents are much more difficult because we'd have to learn each word on a case-by-case basis because they don't correspond to the same lexical sets in our dialects (the same reason why hollywood actors never get the bad-lad thing right). Yeah, I'm with you on the dictionaries for that reason. Why give less information when you could give more? I like how IPA of English words is usually shown on Wikipedia, with as many cross-dialectic distinctions shown as possible (lacking marginal things like bad-lad and assuming a universal fern-fir-fur merger).

I personally don't care a bit if people such as actors who are not native American English speakers imitate NAE with the cot-caught merger. My pet peeve is with dictionaries, because by representing NAE with the cot-caught merger you are ignoring the fact that roughly half of NAE-speakers are cot-caught unmerged, and as you say, why give less when you can give more?

[Travis B.]

I mean, I'd personally prefer it if non-Australians did not assume or act as if Australian English lacked the bad-lad split - Hollywood actors doing Australian accents is excruciating

The problem with BAD-LAD for NAE-speakers is that NAE-speakers today, except for maybe some very elderly people, lack phonemic vowel length, which makes it hard for them to imitate a dialect with it. Rather, as I have commented on very many times here, they have allophonic vowel length, and to my knowledge are largely unaware of it. Likewise, they either don't hear vowel length or they hear it as part of the consonants, combined with other things such as glottalization and aspiration, to make up their perception of fortisness versus lenisness.

[Travis B.]

I know it's easier, but I really would prefer it personally if non-Americans did not assume or act as if standard American English were cot-caught merged. Dictionaries of "American English" which have said merger are a personal pet peeve of mine, as of course you can trivially derive said merger from the lack thereof but not vice versa.

Given that you described this as a pet peeve, I don't know how seriously I should take it, but what does "not assume or act as if" even mean in this context? That it annoys you if someone who isn't a native speaker of American English chooses to adopt or imitate an accent with the merger?

In my opinion, it's hardly more useful to state that "standard American English lacks the cot-caught merger"; or at least, it seems to me that is only true if you use a definition of "standard American English" that refers to a relatively irrelevant theoretical construct (the archetypal, perhaps mythical "General American" of about a century ago). My experience has been that the presence or absence of the merger is largely ignored—at least by those who have it, and we're hardly a negligible minority. I have a hard time imagining a situation where a speaker who has the merger by nature would feel the need to consciously adopt a pronunciation without it when speaking on television/radio/etc.

There's an argument to be made that "standard American English" is just a fairly incoherent and unhelpful concept, but if we choose to speak about it as a thing, I'd consider it most accurate to say that it includes varieties without and with the cot-caught merger. Of course it can be helpful to give unmerged pronunciations in dictionaries, but it isn't all upside, since there is also the possibility of causing confusion for speakers to whom it isn't relevant (this is a stronger consideration for distinctions like horse vs. hoarse or wine vs. whine that are now much less widespread, but could also be argued to belong or have belonged to "General American" or "standard American English").

I was speaking about dictionaries, as mentioned in my previous response to Imralu. I really don't care how people imitate NAE varieties, but it drives me up the wall when I look up a word in a dictionary and see LOT as merged with THOUGHT for the reasons mentioned earlier. To me a General American that is cot-caught unmerged is more inclusive than a cot-caught merged GA, because you can always derive the latter from the former, but not vice versa.

[Imralu]

[...]non-Americans[...]

Given that you described this as a pet peeve, I don't know how seriously I should take it, but what does "not assume or act as if" even mean in this context? That it annoys you if someone who isn't a native speaker of American English chooses to adopt or imitate an accent with the merger?[...]

I was speaking about dictionaries, as mentioned in my previous response to Imralu.[...]

You didn't express that very clearly. In fact, you said "non-Americans" (who, I'm assuming, are not the main authors of dictionaries of American English), which made it seem as though you feel entitled to the world knowing the ins and outs of American English, hence my comments about misconceptions of Australian English. But yeah, I can also see how you might have said that as an effect of the dictionary issue. Also, honestly, I'm more likely to look at a non-American dictionary to see pronunciation in American English because American dictionaries nearly all use awful ad hoc pronunciation guides rather than IPA, so you might be right where it comes to, say, Collins dictionaries mentioning AmE pronunciations, but I see the same issues from the occasional American dictionary that gives British pronunciations as well, but fail to recognise that there is a distinction between /ə/ and /ʌ/ in RP (e.g. "unkind" is not /ənˈkaɪnd/ but /ʌnˈkaɪnd/.)

Nope, they're both NORTH/FORCE

I wonder if there's any dialect that distinguishes them by one of them having NORTH and the other having FORCE.

only a small set of words with ⟨orV⟩ in them are LOT in most NAE varieties, such as tomorrow, sorrow, and for many, sorry (but I personally have NORTH/FORCE in sorry). The exceptions are Canadian English varieties, which tend to have NORTH/FORCE in these words across the board, and some East Coast varieties, particularly amongst older speakers, where words such as Florida also have LOT.

Yeah, I tend to generally assume that you <or> is NORTH/FORCE in US English, because using the LOT/PALM vowel before an <r> (assuming the FATHER-BOTHER merger) is going to lead to it merging with START (or are there varieties where the vowel quality of PALM and START is different and they're not just distinguished by the /r/?)

The problem with BAD-LAD for NAE-speakers is that NAE-speakers today, except for maybe some very elderly people, lack phonemic vowel length, which makes it hard for them to imitate a dialect with it. Rather, as I have commented on very many times here, they have allophonic vowel length, and to my knowledge are largely unaware of it. Likewise, they either don't hear vowel length or they hear it as part of the consonants, combined with other things such as glottalization and aspiration, to make up their perception of fortisness versus lenisness.

Yeah, I think even more generally though, regardless of the presence or absence of phonemic vowel length, it's just generally hard to imitate accents where there are splits you don't have or absences of mergers that you have in your own dialect.

I had a friend years ago who was so terrible at imitating an American accent it was funny. Like, because we merge NORTH-FORCE-THOUGHT, she'd just do either one with any word. For us, "talk" and "torque" are homophones, so she'd say "talk" with /r/. It doesn't immediately stand out as wrong to us like it would to someone with a rhotic accent either.

To do an American accent, we have to think about the spelling to separate things like "talk" and "torque", which takes a bit of energy. It's not as simple as just changing the pronunciation of our own phonemes. In the case of adding rhotacism though, it's fairly simple because the spelling is almost always a good indication (aside from a few weird exceptions like "colonel" and "comfortable"), but this only works for guessing pronunciation of US varieties with the COT-CAUGHT+FATHER-BOTHER merger because the vowel qualities can be unpredictable otherwise. Also, the HURRY-FURRY merger and the ORAL-AURAL merger (not sure if there's another name for that) are easy to apply as a widespread rule, so mostly, we just have to apply rhotacism according to spelling and the rest falls into place because we tend to make more distinctions, but with anything that's distinguished in another dialect where there is no orthographic clue, like the BAD-LAD split, it's much harder because you have to learn for each word. There are some rules and tendencies though. Ablaut for past tense always results in the short one (LAD). The distinction is only really possible in front of voiced consonants and even then, only a few of them. LAD is always used before /ŋ/. Monosyllabic words ending with /d/ are probably the most unpredictable and we end up with a lot of minimal pairs based on morpheme boundaries.

/ˈmæ̆nɪŋ/ Manning (surname)
/ˈmæːnɪŋ/ manning (verb)

/ˈbæ̆nə/ banner (long flag)
/ˈbæːnə/ banner (one who bans)

/ˈtæ̆nə/ Tanner (surname, North-American given name)
/ˈtæːnə/ tanner (one who tans)

/ˈdæ̆ni/ "Danny" (short form of Daniel)
/ˈdæːni/ "Dan-y" (informal adjective derived from "Dan")

/ˈplæ̆nət/ "planet"
/ˈplæːnət/ "plan it"

/spæ̆n/ "span" (variant form of past tense of "spin")
/spæːn/ "span" (gap, expanse)

The word "ass" is a funny one because you'll occasionally hear it with /æ̆/ as a pretty archaic word for "donkey", but mostly we encounter it as an Americanism equivalent to our word "arse", but then it's got /æː/, which we don't natively have before unvoiced consonants, so it sounds kind of awkward, like you switch to an American accent to say "Jackass" /ˈdʒæ̆kˌæːs/, so some people say it with /æ̆/ or with /ɑː/, which other people will find sounds weird, but it's just adapting it to our own phonology.

Anyway, my point was that, even if you learn these rules and tendencies, when it comes to monosyllabic words ending with /d/ and their derivatives, it's unpredictable and it's hard to even find a dictionary that indicates which words:

LAD: add, ad, bade, Brad, cad, Chad, clad, dad, fad, gad, grad, had, nad, pad, plaid, rad, Novy Sad, tad, trad, Vlad, Thad
BAD: bad, glad, mad, sad

In my teaching career, I've mostly taught from textbooks that focus on RP, and with RP it's very easy for me to know what phoneme is used in each word because the distinctions almost always correspond to my own, although I also have the BAD-LAD split and "gone" has a unique vowel, /ɒː/, a longer version of the LOT/CLOTH vowel. The only things that are harder for me are CURE, because I lack it, and unstressed /ɪ/, because unstressed /ɪ/ and /ə/ are mostly conditioned by the following consonant for me. CURE is pretty easy for me to work out when it occurs based on the spelling. It tends to either turn into a bisyllabic GOOSE+COMMA/LETTER sequence (as in "tour", "cure", "pure") or into the THOUGHT/NORTH/FORCE vowel (as in "poor"). The only time it's really tripped me up are in words where there's a vowel after it as in Europe, because I say that with GOOSE. The /r/ is just at the beginning of the next syllable and doesn't affect the vowel for me, so I was surprised to see that Europe is has the CURE vowel in RP, not GOOSE, but there are relatively few words like that and now I assume, if I have GOOSE+/r/, it's probably CURE in RP. The unstressed KIT thing is difficult though because there are so many words with that and for a while, I just kind of assumed that <e> or <i> meant KIT rather than COMMA/LETTER (words with -ed and -es for example), but then I remember seeing the word "cousin" in IPA in a textbook I was using and it had a schwa and I just thought "Fuck. How am I supposed to know?" So, like, I couldn't tell you if RP has KIT or COMMA/LETTER in the second syllable of "chocolate" and now I'm at the point where I also don't care. If a student asks me how it is pronounced, I give them the Australian pronunciation with /ə/ and tell them I'm not sure of whether it's got /ɪ/ or /ə/ in other accents. We can look it up if they really want to know, but the main thing is that it's not [ˈtʃokoleɪt] in native English. (Just looked it up and it's either one in RP, but I generally just don't even hear the difference. Before /t/, unstressed vowels are universally schwa for me, not KIT.

I'd also really love it if people from the US stopped insisting that [ʉ̯] is /r/

What does this refer to?

Americans insisting that Australians say "no" as "nawrrrrr". (There is an accent within Australia that we make fun of for having a kind of "r" pronunciation in a few vowels.) The majority of us say the GOAT vowel as something like [ɐʉ̯] or [əʉ̯] or with a real bogan accent, it sounds like it goes towards [ay̯], and Americans will hear that and imitate it with something like [ɑɻːːː], which sounds completely different (even from the weird Australian accents that have a weird r-like sound to a few vowels)

[bradrn]

/ˈmæ̆nɪŋ/ Manning (surname)
/ˈmæːnɪŋ/ manning (verb)

/ˈbæ̆nə/ banner (long flag)
/ˈbæːnə/ banner (one who bans)

/ˈtæ̆nə/ Tanner (surname, North-American given name)
/ˈtæːnə/ tanner (one who tans)

/ˈdæ̆ni/ "Danny" (short form of Daniel)
/ˈdæːni/ "Dan-y" (informal adjective derived from "Dan")

/ˈplæ̆nət/ "planet"
/ˈplæːnət/ "plan it"

/spæ̆n/ "span" (variant form of past tense of "spin")
/spæːn/ "span" (gap, expanse)

…

LAD: add, ad, bade, Brad, cad, Chad, clad, dad, fad, gad, grad, had, nad, pad, plaid, rad, Novy Sad, tad, trad, Vlad, Thad
BAD: bad, glad, mad, sad

The BAD–LAD split is an interesting one for me… I definitely have a distinction between /æ æː/, but quite a different one. For one thing, my LAD is long, just like BAD, so I can’t really call it a BAD–LAD split to begin with. For the other words which you mention and which I use in my speech:

/æː/ — bad, glad, mad, sad, lad, add, Brad, dad, pad
/æ/ — ad, clad, fad, grad, had, tad, Vlad

So I have a minimal pair between add and ad, which you don’t have. (Also adder and adder — the snake has a short vowel.) On the other hand, I lack your minimal pair for span and possibly banner. I also find ass to be a bit weird to say. In general, my best guess of my overall accent is South African English which has been somewhat Australianised, so I guess this distribution is consistent with that hypothesis.

[Darren]

My BAD-LAD split matches Imralu's pretty closely, except Chad, gad, clad and Brad have BAD (and I think maybe pad does as well, at least for some senses). Other minimal pairs are "can (auxiliary)" /kæn/ vs. "(tin) can" /kæːn/, "Nan" (grandma) /næn/ vs. "Nairne" (place name) /næːn/, "CAD" (computer-aided design) /kæd/ vs "cad" /kæːd/. And MOUTH is [æː] before coda /l/, hence cowl [kʰæːɫʷ] vs. Cal [ˈkʰæɫʷ]. Similar to this is SHOWER [æˑo̯ə̯ ~ æˑə̯ ~ æː] (except in some cases where it's bisyllabic like coward [ˈkæːwəd]), so in a strong accent you might have lad~glowered as a near-minimal pair (which would very neatly provide an open syllable /æː/ with intrusive /ɹ/ like all the other long vowels).

/æ ~ æː/ is definitely not as phonemic as all the other long vowel contrasts (/i ʊ e ə a/ vs. /ɪː ʊː eː əː aː/); none of the other vowels show any length variability and the pair members arise from mostly unrelated sources. There's more of a spectrum for /æ/-lengthening, with some words being securely in one camp and others quite variable:

Always short: CAD, can, had, Vlad, Nairne, Cairns, lad, dad, ad(d), span (< spin), can (auxiliary), all /æ/ with codas other than /b d g m n/
Either sounds alright, but short is probably better: pad, trad, Nan
Either sounds alright, but long is probably better: cad, Chad, clad, Brad
Always long: mad, sad, bad, bag, cab, can (tin), span (bridge), clan, anything with /æm/, all /æo̯l/
Both sound ridiculous: ass (I just say arse; dumbarse and arsehole are ok but jackarse sounds weird)

[Nortaneous]

(or are there varieties where the vowel quality of PALM and START is different and they're not just distinguished by the /r/?)

Yes - for me the onset of START is closer to THOUGHT than to PALM.

Anyway, my point was that, even if you learn these rules and tendencies, when it comes to monosyllabic words ending with /d/ and their derivatives, it's unpredictable and it's hard to even find a dictionary that indicates which words:

LAD: add, ad, bade, Brad, cad, Chad, clad, dad, fad, gad, grad, had, nad, pad, plaid, rad, Novy Sad, tad, trad, Vlad, Thad
BAD: bad, glad, mad, sad

Isn't this the same distribution as the æ-eə split in the US East Coast dialects that have it?

[Man in Space]

Very, though “Welp” often connotes some flavor of resignation or reluctance.

[Travis B.]

(or are there varieties where the vowel quality of PALM and START is different and they're not just distinguished by the /r/?)

Yes - for me the onset of START is closer to THOUGHT than to PALM.

To me START involves conditional raising (e.g. star has [ɑʁˤ], but start has [ʌʁˤ] - properly this involves a phonemic split because Martha, farce, parse, and scarf (n.) generally have [ɑʁˤ] while hearth, marsh, barf, and scarf (v.) normally have [ʌʁˤ], but parse and scarf (n.) can go the other way), while I have the discussed split in LOT/PALM between [a] and [ɑ] (e.g. father, lot, pasta and most LOT words have [a], but palm, wok, rock, hot, aqua, quad, the first syllable of Guatemala, etc. have [ɑ]).