While it’s still far too early for me to have results from my (audio) dialect survey for African American English Accents, I recently got a nice replication of some previous findings. I was at a computational sociolinguistics workshop a few weeks ago, and was graciously allowed access to some 2013 data, to compliment my 2015 work. Anyway, looking at a million tweets, I found the same pattern — South to North — of use of sholl for “sure”. (This is comparing all spelling variants, so sholl, shol, showl, showll, shole, etc. versus sure, shurr, shur, etc.)

Using a Getis-Ord Gi* statistic for hotspot analysis, much the same pattern as I’d argued for in 2015 appears in this different dataset. Note, however, that I did play fast and loose with the K nearest-neighbors, and this is a very preliminary analysis. I’m also still working out how best to represent the “envelope of variation” — so this is an analysis where I’m comparing sholl to sure as a proportion of total tweets.

As I’ve written about elsewhere, this feature and others do not pattern at all like regional variation in (white) English described in the Atlas of North American English.

-----