This question is about speakers without the cot-caught merger (so, speakers who pronounce words such as “lot,” “cot,” “swat" with a distinct vowel from words such as “thought,” “caught,” “water.”) I’ll use broad phonemic transcriptions of the British vowel in “lot” as /ɒ/, the American vowel in “lot” as /ɑ/, and the vowel in “thought” for any speaker as /ɔ/.
Usually, the British phoneme /ɒ/ corresponds to the American phoneme /ɑ/; this is the definition of the “LOT” lexical set. There are known patterns of exceptions where the American phoneme may be /ɔ/ instead, such as before /r/ (where Americans have /ɑ/ or /ɔ/ depending on the word, with some variation also by dialect) and before the phonemes /f/, /θ/, /s/, /ŋ/, /ɡ/, and a few unpatterned exceptions for some speakers such as “chocolate,” “on” and “gone.” Generally, these words are called the “CLOTH” lexical set. I read about this sound change in the following places: Phonological history of English low back vowels: Lot–cloth split (Wikipedia), The Cloth Set (Dialect Blog, by Ben Trawick-Smith)
Some other words such as “salt," "Australia," "sausage," "laurel" and "cauliflower," which originally had /ɔ/, have /ɒ/ in current British English (which dictionaries mark as either being in variation with /ɔ/, or invariable) due to vowel shortening that is apparently attested as far back as 1800. Most of these words originally had /ɔl/ or /ɔs/ before a consonant. For these words, dictionaries seem to only record American pronunciations with /ɔ/; I don't know if this should be considered a retention (lack of the shortening change /ɔ/ > /ɒ/), a restoration (possibly based on the spelling and/or due to CLOTH lengthening), or some mix of both. I read about this sound change at the following places: "scolding water" (John Wells’s phonetic blog), and the article The History of [ɔː]: Is There Regular Orthographically Conditioned Sound Change?, by Piotr Gąsiorowski
However, there are at least two other environments where British /ɒ/ may correspond to American /ɔ/ (I say "may" because dictionaries generally also list /ɑl/ as an option in the American pronunciation of the following words):
- Words that originally had /ɒl/ at the end of a word or before a consonant (such words are somewhat rare, since most inherited words that would have this pattern seem to have shifted to /oʊl/). For example, “alcohol,” “golf,” “solve,” “dolphin," "doll." For the standard accents transcribed in dictionaries, the normal reflex of intervocalic /ɒl/ in American English is /ɑl/ (in words like "anabolic") but a few words may be pronounced with intervocalic /ɔl/ by some speakers due to analogy, such as ""alcoholic" (which is related to "alcohol") and "workaholic" (which is based on "alcoholic"). However, I did find a very interesting forum post by one speaker who describes having /ɔ/ (phonetically [ɒ]) in THOUGHT words, CLOTH words, and any words where the vowel is followed by "l" (including intervocalically), but [ɑ~a] in LOT words. The speaker says:
I have a CALLER-COLLAR merger, but COT-CAUGHT, HOCK-HAWK, and DON-DAWN are all unmerged.
CALLER-COLLAR has [Q].
COT, HOCK, and DON have [a].
CAUGHT, HAWK, and DAWN have [Q].
–a more original name, on the Antimoon Forum
- Words that in Britain have /ɒʃ/, such as “wash,” “swash,” “squash,” “quash,” “slosh.”
Aside from the above forum post, I haven't read about this sound change anywhere; I've had to infer it from entries that transcribe non-cot-caught-merged American pronunciation in the Merriam-Webster Dictionary, American Heritage Dictionary, and Oxford Advanced Learners Dictionary. The pronunciations these dictionaries record do not suggest a conditional merger or complete loss of distinction in this context (like the COLLAR-COLLER speaker above, or like the DON–DAWN merger that is described in the Wikipedia article) since some words with the same phonetic environment are transcribed differently ("alcohol" is mainly transcribed with /ɔl/, while "loll" is only transcribed with /ɑl/; both /ɑl/ and /ɔl/ are listed for "golf," while only /ɑl/ and /oʊl/ are listed for "olfaction"). If this is accurate, it suggests to me that the change is somewhat well-established and no longer automatic for all speakers, although maybe the opposite is true, that it is still in progress and in the process of diffusing to all words with the relevant environment.
There is a list I found, taken from John Wells' Accents of English: Introduction Chapter 2, that mentions that "wash" is in the CLOTH lexical set. But I haven’t found any description of CLOTH that mentions words like "doll" and "solve," even though they appear to be in the CLOTH set for enough speakers to get transcribed in dictionaries. (In fact, that list explicitly says that "doll" and "solve" are in the LOT set.)
What is a good resource that gives a more complete description of the CLOTH lexical set, including all of the possible conditioning environments (such as the ones above, and perhaps more that I am unaware of) and its development over time? I would like to read it. Obviously it is more convenient for me if it is online and free to access, but if an offline or for-pay text provides a better description, I'd prefer that.
Note: I asked a similar question earlier on the English Language Stack Exchange, and received many helpful comments that allowed me to clarify my thought process, but no citations or pointers to existing descriptions of this change that have been published by linguists. It was suggested that I might have better luck here. I hope that's all right; I have completely re-written the question and hopefully it's clear what I want to learn, and how this relates to linguistics. You can look at that question though if you want to see a more complete list of affected words.