What source explains the different pronunciations of "hol" in "alcohol" and "hollow"?

Question

According to Merriam-Webster, the pronunciation of alcohol is "ˈal-kə-ˌhȯl" while the pronunciation of hollow is "ˈhä-(ˌ)lō." Why are they pronounced with different vowels? I think I've figured out the reason (my explanation is further below) but I haven't been able to find any source to confirm my guess.

Background explanation of the two sounds /ɑ/ and /ɔ/ in American English

In case you're wondering how they could possibly be pronounced differently, or what Merriam-Webster's pronunciation symbols mean, here is some background information. For many American English speakers, the words "cot" and "caught" don't rhyme because they have different vowels. (For many others, they do: speakers like this have merged the vowels.) In this question, I'll use the International Phonetic Alphabet (IPA) and follow the usual convention of representing the North American vowel in words like "cot" and "lot" as /ɑ/ and the vowel in words like "caught" and "thought" as /ɔ/ (for speakers who have not merged this vowel with with /ɑ/). Standard British English also makes this distinction, but uses slightly different vowels; the usual IPA transcription for the British "cot"/"lot" vowel is therefore slightly different (it's /ɒ/) and the British "thought" vowel is standardly transcribed as /ɔː/ (with a vowel length marker ː). Anyway, Merriam-Webster uses a different transcription system that represents the vowel in cot (IPA /ɑ/) as \ä\ and the one in caught (IPA /ɔ/) as \ȯ\.

Generally speaking, the "lot" vowel, /ɑ/ (or for British speakers, /ɒ/), is used in words spelled with "short o" (like don, cot, body), and the "thought" vowel, /ɔ/, is used in words spelled with "aw" or "au" (such as dawn, caught, bawdy). There are several other more minor spelling patterns that you can generally find described in works on English phonology.

Background explanation of the lot-cloth split

The most important exception to the generalization I made above is that certain specific words spelled with "short o" in American English are pronounced with /ɔ/ instead of /ɑ/. This group of words is exemplified by the word "cloth" /klɔθ/, and is the result of a historical sound change, the LOT-CLOTH split, that has gone to completion in American English, but not in British English. Generally, the sound change occured in specific, predictable contexts.

Wikipedia says:

The lengthening and raising generally happened before the fricatives /f/, /θ/ and /s/. In American English the raising was extended to the environment before /ŋ/ and /ɡ/, and in a few words before /k/ as well, giving pronunciations like /lɔːŋ/ for long, /dɔːɡ/ for dog and /ˈtʃɔːklᵻt/ for chocolate.

All the sources I have found agree with this: they say that this change affected "short o" before the sounds f, th, s, ng, g, and a handful of words with n (gone and on).

An odd spelling pattern: ol at the end of a word

I have not found any source that says that words spelled with "ol" are included in the "cloth" set. As you can see, Wikipedia doesn't mention words like "alcohol" or "golf," and even though I've done some Google searches to look for more information on the "cloth" set, I didn't find anything that discusses "ol" words.

The dictionaries that I've looked at only list /ɒl/ in the British pronunciations of these words; I would expect this to correspond to American English /ɑl/ (as it does in the words dollar /ˈdɑlər/ and tolerate /ˈtɑləˌreɪt/).

However, I have found many words spelled with "ol," such as alcohol, parasol and aerosol, that dictionaries say are pronounced in American English with /ɔl/ (as if they were spelled with "awl").

What I think the reason is

I believe I have identified the condition for this sound change, thanks to the helpful comments from people like Peter Shor who described their pronunciations. The /ɔl/ pronunciation seems to occur mainly when "ol" is at the end of a word or before a consonant. It doesn't seem to have anything to do with the preceding consonant: complare golf, alcohol, aerosol (all words that may be pronounced with /ɔl/), and golliwog, hollow, solid (words that as far as I can tell are always pronounced with /ɑl/).

Pronunciations with /ɔl/ are listed for a few words where the "l" is between vowels, such as alcoholic and cytosolic, but when this occurs it always seems to be due to analogy from the above set of words (Merriam Webster only lists a pronunciation with /ɑl/ for melancholic, which does not have a corresponding noun melanchol to influence its pronunciation).

What I still would like to know

1. Is there any source that mentions the existence of this sound change before /l/?
2. When did words like "alcohol" start to be pronounced with the sound /ɔl/? Some parts of the LOT-CLOTH split are attested in earlier British English (like "orphan"/"often"); are there any parallel attested cases of British /ɔl/ instead of /ɒl/? I'd guess not, since as far as I know lengthening before /ŋ/ and /ɡ/ never occured in British dialects.
3. Is the pronunciation with /ɔl/ currently universal (for at least some words) among American speakers without the cot-caught merger, or do some of them pronounce "ol" as /ɑl/ exclusively? In other words, are there any American English speakers for whom "alcohol" does not rhyme with "all"? Are there any speakers that use /ɑl/ in some of these words, and /ɔl/ in others? Merriam-Webster only records /ɔl/ for "alcohol," but for many other words like this it records both /ɑl/ and /ɔl/ as alternate pronunciations. It also only records /ɔl/ for words like "awl." If we assume this is 100% accurate, it would mean there are some speakers that have /ɔl/ in "alcohol" but /ɑl/ in other words, such as "alcoholic." However, I'm not sure that it is accurate, and in any case, it still doesn't clearly describe the overall pattern of variation that would be expected from a single speaker.

As a bonus, I would appreciate if an answer addressed the phonetic reason for this sound change. Is it due to the lengthening process that seems to have created most of the rest of the CLOTH set? Or is it a separate change caused by the rounding/backing effect of "dark l"? Ben Trawick-Smith has a post about "The Cloth Set" on his dialect blog where he suggests that velarization may have been the reason "o" is pronounced /ɔ/ before some words with /ŋ/, /ɡ/, and /n/; since "dark l" is velarized, this seems like it would also be applicable here.

Examples

Here is a list of words spelled with "ol" that are, or can be, pronounced with /ɔl/. I consulted online versions of the Merriam-Webster dictionary (MW), the Oxford Advanced Learner's Dictionary (OALD), and the American Heritage Dictionary (AHD):

Two pronunciations /ɔl/, /oʊl/ listed in MW, AHD

thymol

Two pronunciations /ɔl/, /oʊl/ listed in MW; only /ɔl/ listed in AHD

menthol

Two pronunciations /ɔl/, /oʊl/ listed in MW; three pronunciations /ɔl/, /oʊl/, /ɑl/ listed in AHD

ethanol
methanol

Two pronunciations /oʊl/, /ɔl/ listed in MW; two pronunciations /ɔl/, /ɑl/ listed in AHD

diol

Only /ɔl/ listed in MW; /ɔl/, /ɑl/ listed in AHD and OALD:

alcohol
alcoholism (pronunciation with \-kə-hə-\ also listed by MW)

Two pronunciations /ɔl/, /ɑl/ listed (MW, AHD, OALD all agree):

alcoholic
workaholic, workaholism
parasol

Two pronunciations listed: /ɔl/, /ɑl/ in MW, AHD and /ɑl/, /ɔl/ in OALD

chocoholic/chocaholic

Two pronunciations listed: /ɑl/, /ɔl/ in MW and /ɔl/, /ɑl/ in AHD, OALD

aerosol

Two pronunciations /ɑl/, /ɔl/ listed (MW, AHD, OALD all agree):

golf (MW says l can be dropped)
solv- in solve, absolve, resolve, solvent, solvency...

Two pronunciations /ɑl/, /ɔl/ listed in MW, OALD; only /ɑl/ listed in AHD:

-volv- in evolve, revolve, involve, devolve...

Two pronunciations /ɑl/, /ɔl/ listed in MW, AHD; only /ɑ/ listed in OALD:

dolphin

Two pronunciations /ɑl/, /ɔl/ listed in MW; only /ɑ/ listed in AHD, OALD:

doll
moll

Two pronunciations /ɑl/, /ɔl/ listed in MW; no entries in OALD

sol (as a unit of currency, or "a fluid colloidal system") 
cytosol, cytosolic
hydrosol

Similar words for which MW only lists /ɑl/

Sol (the Roman god) (AHD lists /ɑl/, /oʊl/)
pol (short for "politician") (AHD lists only /ɔl/)

For comparison:

Similar words for which dictionaries list only /ɑl/, not /ɔl/ (MW, AHD, OALD all agree)

col
loll
hydrosolic (not in OALD)

melancholic
metabolic
vitriolic 
diastolic

Similar words MW, AHD list with /ɑl/,/oʊl/ (but not /ɔl/)

olfaction, olfactory

Answer 1

Given all your examples of supposedly rhyming words (they all rhyme for me a GenAmE speaker) having different possible pronunciations, both within a dictionary and between dictionaries, leads me to believe primarily not that there is some variation in pronunciation in the GenAmE population (though there very well may be variation (but consistent) between regional accents), but rather that these dictionaries are not scientifically consistent within themselves.

Either you have discovered an substantive inconsistency among these entries or these dictionary makers have performed research on all these pronunciation as part of their preparation but are not printing the outcomes/final results of the data leading them to make distinctions that are not meaningful for actual pronunciation or hearing.

You may want to construct a diplomatically worded letter to the editor in order to discover the reasoning for these pronunciations. I wonder if it would help to show one dictionary what the other dictionaries say?

Answer 2

1. Like you I cannot find any references as to why there is a distinct pronunciation differences around the "o." All I can figure is that it is simply a mutation that has simply "stuck." More than likely it has something to do with the reason why American accents are considerably different than British accents the speed in which they diverted.

2. This is more about the letters around it. The letter "h" seems to have problems in many languages, and especially so in Romance languages. The French have been ignoring it for centuries! Both in English and in German it is generally pronounced as a breathless letter. Consequently it tends to make the next phonic sound "longer." For example Hard and Card. The "h" makes Hard take longer to say and you can really hear the "a" appose to Card which can sound more like "Crd." Then you have the "l." This adds to the already overly stressed "o" to make sound even more like "awl." Take words like all, null, and doll, they all have an "wl" sound to them when pronounced. So overall the "dark l" is what is causing it, but with significant help from a breathless "h" to make it a definite "awl."

3. I'm going to combine #3 and #4 since they have the same answer. British pronunciation in dictionaries use the /ɒl/ and no variation and American pronunciations use mostly /ɔl/ to some variation. I would assume that would mean that it is fairly resent that the "common" pronunciation is "awl" but it has be significantly different from the British since the deviation first start. This will come down the how the speaker pronounces the "h" as to how close to "a" the "o" will become, causing the deviation in the phonetic key.
4. We are back to the vowel + dark "l" combination again. With the syllables separating the "l"s in the American pronunciation you end up with "col" and "mol". Both are pronounced with the "wl" but with the lack of "h" makes it sound like /ɑl/.

Overall Americans pronounce "ol" with an accent that is very similar to "awl." The differences being literally technicality, which can be seen in the dictionaries. All of them are using a different phonetic alphabet meaning the exact characters' pronunciation equivalents are up to debate. Considering your MW dictionary uses a different one than the one I found online at their website. As for refences to the British pronunciation I used dictionary.com. Which shows every "ol" combination as /ɒl/.