Counting and coding the past: Circumscribing the variable context in quantitative analyses of past inflection

Abstract Accurate circumscription of the variable context is crucial to any quantitative analysis of linguistic variation. Investigations of past inflection in African American Vernacular English and Caribbean English creoles thus generally include a more or less detailed section concerning the inclusion or exclusion of particular forms; the theoretical grounds on which these decisions are made, however, are not always spelled out. Consequently, there still does not seem to be agreement on what precisely constitutes the envelope of variation in such investigations—a fact that not only complicates data extraction and analysis but also hampers cross-variety comparisons. This article summarizes and evaluates previous definitions of the scope and relevant contexts of the variable (ed), providing internal (linguistic) argumentation supporting or contraindicating the inclusion or exclusion of particular tokens. My data stem from a larger study of past temporal reference in the urban variety of Bahamian Creole English (Hackert, Stephanie. [2004]. Urban Bahamian Creole: System and variation . Amsterdam: Benjamins), an intermediate creole with close historical links with Gullah as well as relations with African American Vernacular English, Trinidadian Creole, and Barbadian.