Miller, Jason Paul. “Mixed sociological alignment and ethnic identity: r-lessness in a Native American community.” Thesis. Raleigh, NC: North Carolina State University, 1996. 84 pages.
Miller examines the levels of /r/ constriction and vocalization among cross-generational, cross-ethnic samples of Robeson County residents. One purpose of his study is to determine whether some speakers might alter their /r/ productions to emulate or to distance themselves from other ethic groups, and whether the socioeconomic hierarchy in the county has linguistic correlates. Miller reviews some key events from Lumbee history to demonstrate that Robeson County shows socioeconomic stratification and that its social dynamics are mainly determined by ethnicity. He briefly reviews the Henry Berry Lowry era, the KuKlux Klan routing of 1958, and the Robesonian hostage-taking of 1988. Some documentation for ethnic tension comes from a 1994 interview of an African-American couple, a white college student, and a Lumbee couple, all conducted as part of North Carolina State University's “North Carolina Language and Life Project” (excerpts provided). Miller then gives an account of the phonetic composition of /r/ and the recent history of the r-less variation in the U.S. and Britain (with theories on why they occurred in various regions). To study preconsonantal and final /r/, Miller used recorded interviews, taken in the informant's home, which were collected as part of the N.C. Language and Life Project. The external constraints of the study were age, ethnicity, and gender. When possible, he chose two members of each gender from each of four age groups and from each ethnicity. He obtained a total of 32 speakers. Internal constraints (which correlated with linguistic production of a certain variant) were syllable stress; whether consonants, vowel, or pause followed the variant; and syllabic positioning of /r/. Miller used transcribers and a spreadsheet to collect 100 tokens of possible /r/ construction from each speaker and then classed them according to the internal and external constraints. He analyzed the data using the Varbrul program. He found that whites favor /r/ constriction, African Americans disfavor pronunciation of /r/, and Lumbee fall between the two. There are significant weightings in the age group categories, however. Miller's Conclusions section proposed reasons for these differences.