Gerard Van Herk
 

Recent scholarly interest in regional variation in contemporary African American English encourages us to investigate the possibility of widespread regional variation at an earlier stage of the language. To date, however, putative regional differences have most often been invoked when discussing method -- to invalidate recently-discovered sources of information on early AAE, as part of a larger and often acrimonious debate on AAE origins. This debate implicitly assumes a model of a single authentic variety from which today’s AAE descended, surrounded by marginal offshoots of limited theoretical interest.
            Building from and addressing some recent sallies in the validation debate (Montgomery 2006, Singler 2007), I suggest how an investigation of regional variation and corpus validation as separate entities might proceed.
            I use findings on verb marking in early written documents and the African American diaspora to propose that regional and social variation was so widespread in early AAE that it would be productive to abandon the single-variety model. I consider the utility of a model in which speakers had (differential) access to a number of (authentic) varieties of AAE; differing social forces over time (especially the intense contact and identity exigencies of the Great Migration) would favor the adoption or retention of particular features, perhaps even those that previously enjoyed limited distribution. Treating region and interlocutor effects as evidence of available informant repertoire, rather than as an opportunity to reject associated findings, shifts the focus of the corpus validation debate towards the issue of data treatment – preservation, transcription, accessibility, and triangulation. In this respect, I offer some tentative suggestions for enriching the utility of existing data sources, including my own.