Urban centers and American English lexical variation

William Kretzschmar

Sat. 4:00-5:40 D

We now have an opportunity to reexamine with new quantitative methods the position of Kurath and McDavid on the relationship of settlement history and the role of colonial cities in the creation of American dialect patterns. It has recently become possible to generate maps for the Linguistic Atlas of the Middle and South Atlantic States (LAMSAS, field work in the 1930s and 1940s) which plot, for every location in the survey area, the probability that a target linguistic feature might have been elicited there at the time of the survey. Maps were created for 115 vocabulary variants, every lexical variant that was elicited from 50 or more of the 1,162 LAMSAS speakers for those 15% of the questions from the LAMSAS survey that have been entered so far into computer storage. The principal benefit of use of these maps is that each map gives a comprehensive analysis of the feature whose distribution it illustrates, and that the set of maps provides comprehensive coverage of the evidence so far available on computer.

Comparison of the maps to Kurath's isoglosses reveals certain problems, and the whole set of maps do not appear to confirm Kurath's regional patterns. The strongest conclusion to be drawn from observation of the entire set of maps is assent to the old saw of dialectologists, "each word has its own history". In order to assess the participation of cities in patterns in the set of variants, I checked the probability of occurrence of each variant in each of 20 cities: Rochester, New York City (NY); Philadelphia, Harrisburg, Pittsburgh (PA); Baltimore (MD); Washington (DC); Charleston (WV); Richmond, Norfolk, Roanoke (VA); Raleigh, Wilmington, Charlotte (NC); Charleston, Columbia, Greenville (SC); and Savannah, Augusta, Atlanta (GA). For more detailed studies I reduced the data set to just those (58) variants which occurred with 75% to 100% probability in at least one city ("marked" forms). Calculations based on the distribution of the variants within the set of variants and between cities confirm earlier studies of the structure of lexical variation (e.g. Kretzschmar, Johnson). Manipulation of lists of marked features by city (in tables similar to those used in implicational scaling) suggest new ideas about relationships between urban centers, especially in terms of the classic work of Kurath and McDavid.