John Nerbonne

University of Groningen

Audio | PowerPoint Presentation in PDF Format

Aggregate Variation in the South in LAMSAS

Summary: We use data from The Linguistic Atlas of the Middle and South Atlantic States (LAMSAS) (Kretzschmar 1994), focusing on the Southern states to provide a characterization of aggregate linguistic variation in the South.

Traditional regional dialect divisions have been based on linguists' selection of a small number of features which are each realized differently in the same various subregions under investigation, and the work of modern variationists has likewise restricted its attention to a small number of features. In the hands of masters this methodology has led to insightful characterizations of the dialect landscape. But the methodology has resisted analytical justification in several points: first, the choice of which features to focus on is not determined theoretically, but only with an eye to the area to be classified; second, every division results in exceptional data, for which no satisfactory treament has been suggested; and third, some groups of linguistic features delineate regions of imperfect overlap, inevitably giving rise to so-called "transitional zones", which suggest that exact overlap of discrete features is a pure foundation for dialect geography.

We use data from the admirably accessible internet archive of The Linguistic Atlas of the Middle and South Atlantic States (LAMSAS) consists of 1162 interviews conducted over a period of thirty years. We focus here on the phonetic variation in LAMSAS, in which 151 pronunciations were elicited, and especially on pronunciations in the Southern part in LAMSAS (Virginia and further south). We seek solutions to the analytical problems noted above through dialectometry, the measure of linguistic differences pioneered especially by Goebl (1984). Following Goebl, we analyze lexical differences at a nominal data level (see Nerbonne and Kleiweg, 2003), and extending Goebl's methods, we measure pronunciation difference not via differences in individual features, but rather via a measure of string distance (Nerbonne et al. 1999). The dialectometric view solves the problems noted above by using a fixed measure of difference and weighing all the evidence available in the dialect atlas, including counterindicating and exceptional data. The key step in dialectometry is from the measurement of individual linguistic variables (the pronunciation of the diphthong /aI/ the lexicalization of the concept 'dragonfly') to the measurement of aggregate differences of varieties. In this dialectometry distinguishes itself from traditional dialectology but also the mainstream in contemporary variationist linguistics.

Besides solving the analytical problems of traditional dialectology, the dialectometric view offers new opportunities for mapping variation, and for addressing questions such as the determinants of variation. We illustrate these in the talk. We also note that dialectometry imposes conditions on data that are difficult to meet. In particular we need comparable data, which requires consistency in data collection techniques. The level of consistency is difficult for a single field worker to maintain, and the LAMSAS materials from South Carolina and Georgia were collected by eight different fieldworkers, who clearly varied in their elicitation techniques. This combination makes the LAMSAS South a difficult target. The talk will sketch a strategy for dealing with such differences that is very much the subject of ongoing work.

References
Goebl, Hans. Dialektometrische Studien: Anhand italoromanischer, räoromanischerund galloromanischer Sprachmaterialien aus
    AIS und ALF. 3 Vol. Tüingen: Max Niemeyer, 1984.
Kretzschmar, William A. (ed.) Handbook of the Linguistic Atlas of the Middle and South Atlantic States, Chicago: The University of
    Chicago Press. 1994.
Nerbonne, John, Wilbert Heeringa, and Peter Kleiweg. Edit Distance and Dialect Proximity. In David Sankoff and Joseph Kruskal
    (eds.) Time Warps, String Edits and Macromolecules: The Theory and Practice of Sequence Comparison, 2nd ed., Stanford:
    CSLI. v-xv. 1999.
Nerbonne, John and Peter Kleiweg, Lexical Variation in LAMSAS, Computers and the Humanities 37(4). Special Issue on
    Computational Methods in Dialectometry edited by John Nerbonne and William A. Kretzschmar. 2003.