Richly Qualitative and Rigorously Quantitative

In 1996, with Tim Frazer, we published an article in which we mapped the geographic boundaries of need + V-en (The car needs washed, The shirts need ironed; see Frazer, Murray, and Simon 1996) in American English. At the time, we conceived of our investigation as a contribution to the ongoing discussion concerning the existence of a distinct Midland dialect in the United States. In the process of data collection, we ended up with provocative material on two related constructions, want + V-en (The baby wants picked up, The cat wants fed) and like + V-en (The dog likes petted, Babies like cuddled) and decided our investigation must include these (Murray and Simon 1999, 2000). During this same period, we also looked at the regional variation in the lexicalized pronunciation of suite 'set of furniture', /swit/ versus /sut/ (Simon and Murray 1999).

We did not realize that what had begun as a set of more or less conventional studies in regional dialectology would have far-reaching implications for dialectological and sociolinguistic methodology, for approaches to the study of American English, and for general linguistic theory. Because of the ways we collected and analyzed our data, we found ourselves calling into question established wisdom regarding regional and social dialect study--questions regarding the significance and use of infrequently occurring forms, regarding contextual and qualitative data, regarding dialect boundaries, and regarding the intersection of syntax, semantics, and pragmatics.

By relying on approaches drawn from several subfields of linguistics--interactional sociolinguistics, sociohistorical linguistics, and formal linguistic analysis--as well as conventional dialectology, we have been able to collect different kinds of relevant data, and we have been able to take full account of those kinds of data. For instance, we examined sources for historical background in conjunction with corpus analysis for those historical periods. We conducted written survey questionnaires and telephone surveys along with posting open-ended questions to targeted listservs and bulletin boards. We applied formal and functional linguistic analyses and then used them to cast light on respondents' own understandings of the item. And we took respondents' expressed perceptions and attitudes about language use as a central point of use. We found that this interdisciplinary approach to collection and analysis is both richly qualitative and rigorously quantitative. And we found that by accounting for all the information collected, we were coherently and systematically addressing important, broad theoretical issues in dialectology, sociolinguistics, and general linguistics.

We have discussed elsewhere at some length how to investigate dialect using combined methodologies (Simon and Murray 1999). Here we want to say that our work has shown how useful it can be to have complementary sets of data, which, besides validating one another, often produce a far more complex regional and social picture of the feature being studied than could be obtained through any one of the sets of data alone. Conventional sorts of dialect surveys and questionnaires certainly have their place in the collection of brief, on-task answers. What we are sure of, though, is that they are not enough and that they are not necessarily the heart of the matter. Locating language study within the specific contexts of language use is crucial to adequate and accurate description and to useful theorizing.

Conventional dialectology and sociolinguistics are wary of phenomena that are not quantitatively significant, but perhaps frequency of occurrence is, at least partly, a false problem. Why not reconceive existing paradigms to accommodate data on variables not quantifiable in the traditional way? It seems clear to us that if most, if not quite all, differences in form are referentially meaningful, the nature of sociolinguistic investigation should be modified. Collecting and analyzing extended interviews, ethnographic participant observation, oral histories, and the like will open dialect study to more types of linguistic phenomena and the specific contexts in which they occur, which is critical to teasing...