ABSTRACT
Many methodological issues can arise when working with hard to find sociolinguistic variables that are not phonological, especially certain morphosyntactic or discourse/pragmatic variables. Some of these difficulties include establishing semantically or functionally equivalent variants, dealing with pragmatic constraints on the variable context, and finding enough tokens for statistical analysis. While some of these issues may ultimately lay with theoretical concerns as to the nature of the linguistic variable itself and its extension to areas other than the phonological, at least one way to partially overcome them is by looking in new places and with new expectations. With hard to study variables like the double modal, constraints on usage may have more to do with pragmatics than is the case for other sociolinguistic variables. Medical consultation corpora such as the Verilogue corpus may thus be quite useful for studying variables like the double modal, which have resisted traditional methods given their low occurrence.
