In Kurath and McDavid's Pronunciation of English in the Atlantic States , the distinction of /hw/ and /w/ in whale and wail, which and witch was shown to be characteristic of the North and the South, but not the Midland. In fact, the southern limit of the /hw/~/w/ distinction bundled tightly with the lexical isogloss separating the North and North Midland through Pennsylvania.
Since the LAMSAS data was gathered, the distinction has rapidly eroded. Map 8 shows only 71 of 587 speakers who maintain it. In this case, "Distinct" includes all those who were heard by the analyst as pronouncing the voiceless bilabial clearly (62 cases) or not quite clearly (9) cases. There were 3 individuals who thought that the pairs were different, but made no distinction in production; they were considered to be merged.
The /hw~w/ distinction is completely absent in New York State, where it was registered quite strongly in the LAMSAS data. It is most clearly retained in two areas of the Eastern States: to a certain extent in Eastern New England (excluding Maine, as in the LAMSAS records), and quite solidly in the Lower South (excluding South Carolina, again echoing the LAMSAS data). It is also quite strongly maintained in Texas, particularly in the Dallas/Forth Worth region.
The scattering of distinct points in the Midwest does not clearly follow the North/North Midland isogloss. There are eight speakers who maintain the distinction north of that line, and nine south of it.