The New England "short o" has been discussed in the literature on American dialects a number of times (see Avis 1961; Kurath -- McDavid 1961: 12). Kurath (1964: 150) introduces this phenomenon as follows. He discusses the loss in American English of the original Middle English distinction between monophthongal Q: and diphthongal ou, and writes:
Only New England [in the USA] preserves the original [ME] distinction, though to a limited extent. Here the old monophthong survives in checked position as a short and fronted mid-back vowel /e/ as in stone, road, coat /sten, red, ket/, contrasting with up-gliding /o/, as in know, grown (but also, e.g., in no, rode). This so-called "New England short o" is somehow related to regional English folk speech.
Avis (1961) further tells us that the heartland of this phenomenon in North America lies, for the USA, in eastern Vermont, New Hampshire, northeastern Massachusetts, and Maine, and in Canada in southwestern New Brunswick.
Kurath (1965 ) asks the interesting question: "Is the survival of contrasting vowels in New England to be attributed to English folk speech?" and answers it as follows: "New England usage in this matter probably derives from English folk speech or from a regional type of Standard British English reflecting folk usage." In a much earlier publication (1928 ) he actually appears to give a more geographically detailed answer to the question when he says: "The population of the seaboard of New England had come for the most part from southeastern counties of England"; and "the shortened vowel of coat, whole, and home is recorded for East Anglia." It is definitely tempting to see a connection. This was certainly also the link I was suggesting myself when I wrote (1974) that there was a "clear resemblance" between the two phenomena in East Anglia and New England.
Further linguistic evidence in favour of Kurath's linkage to northern East Anglia is as follows. In the case of both East Anglia and New England, the phenomenon of shortening is confined to items descended from Middle English Q:; it never occurs in either area in words such as grown, soul. Both features, moreover, are recessive, and both vary widely in their incidence from word to word, style to style, and speaker to speaker.
The East Anglian facts are as follows. The vowel of English labelled by Wells (1982) as the GOAT vowel has, as in New England, maintained two counterparts in the vowel system of the dialects of northern East Anglia, that is Norfolk and northern Suffolk. Paralleling a now vestigial distinction in the front vowel system between the sets of made and maid, corresponding to the distinction between the Middle English monophthong and diphthong, there is in northern East Anglia a similar contrast in the back vowel system which, however, is by no means vestigial (see Trudgill 1998 on why the one distinction has survived longer than the other). The distinction is between /u:/ = [uu], descended from ME Q:, and foul = [eu], descended from ME ou. Thus pairs such as moan-mown, road-rowed, nose-knows, sole-soul are not homophonous.
The important thing to notice for our purposes is that the FOOT vowel /u/ was much more frequent in the older East Anglian dialect than in General English (in the sense of Wells 1982). There has been a strong tendency in East Anglia for the /u:/ descended from ME Q: to be shortened to /u/ in closed syllables. Thus road can rhyme with good, and we find pronunciations such as toad, home, stone, whole, coat /tud, hum, stun, kut/. Distribution is unpredictable: /u/ does not occur, at least in current dialects, in foam, load, moan, coal, vote, for example.
The extent of this East Anglian "short o" phenomenon is indicated in the work of a number of writers. Kokeritz (1932), for example, writing about the dialects of northern Suffolk, lists the following items as having the same vowel as pull:
boast, boat, bone, choke, cloak, clover, coach, coast, coat, don't, folk, goat, hole, home, hope, load, loaf moat, most, oak, oath, oats, over, poach, pole, post, road, rope, smoke, stone, toad, whole, wholly
Lowman's records (see Trudgill 1974) also show a large number of examples of the "short o", although (possibly incorrect) transcriptions such as [ston] stone make it unclear as to whether he regards such words as having a vowel identical to that of foot. …