-
1. Collection and Categorization of the Data
- The University of Alabama Press
- Chapter
- Additional Information
1 Collection and Categorization of the Data The conclusions of this project are based on an examination of two sets of data: one collected in the late 1930s as part of the Linguistic Atlas of the Middle and South Atlantic States (LAMSAS), and one collected by the present researcher in 1990. These sets include pairs of informants matched according to personal characteristics and living in the same Georgia, South Carolina, and North Carolina communities. This chapter will describe the survey design and methodology under two major headings: linguistic variables and demographic variables. The first covers the items chosen for the analysis and the ways they were elicited and recorded; the second discusses the choice of informants and the ways they were categorized. The terms question, survey item, and linguistic variable are used interchangeably to refer to a cue designed to elicit an array ofwords that designate more or less the same object. Likewise, word, synonym, tenn, response, and especially, variant are used to designate the members of a set of verbal responses to each cue. Social variables refer to age, sex, race, and education as grouping devices for categorizing speakers, while the regional variables are rurality and region. '!he Linguistic Variables and the Interview The 150 linguistic variables chosen for the analysis are all lexical items. They form a subset of items included in the LAMSAS questionnaire , and cover topics such as farm animals, household items, clothing, food, greetings, weather, and other everyday terms. The responses for each question have been grouped into eleven databases according to their typical order in the interview and their topic or semantic field. These are listed in Table 1. 6 Data Table 1: Questionnaire topics 1) rooms, furniture 2) milking-related terms, other calls to animals 3) outbuildings, architectural features, kitchen utensils 4) modes of transportation, tools 5) domestic animals and sounds they make, weather terms 6) com, breads, snacks 7) meats, fruits, vegetables 8) insects, wild animals, plants 9) kinship terms, items related to childbirth 10) marriage, death and illness 11) school-related terms, holidays, clothing Items chosen from the LAMSAS worksheets fit the following guidelines: (1) If the compilation of linguistic atlas worksheets by Davis et aI. (1969) listed several possible responses (e.g. fire dogs, dog irons, and andirons) to the question, it was assumed the question would be productive in eliciting synonyms, thus providing a range of lexical variation; (2) Some items that were less productive of lexical variation were included because it seemed likely that they would have changed in distribution by the late-twentieth century, e.g., words like clabber (thick sour milk), and singletree and other words referring to aspects of horsedrawn transportation. The amount of change documented in this study is thus not representative of the proportion of items likely to change in the lexicon as a whole; and (3) A number of items were chosen because they were analyzed by Hans Kurath in his Word Geography of the Eastern United States (Word Geography) (1949). This gives the added aspect to the study of verifying by computer the distributions of lexical items claimed by Kurath-and long assumed by others-to reveal the major dialect boundaries of the United States. Following these guidelines, 240 questions were considered for inclusion because they were likely to produce evidence of either variation or change, or because they were responses that could be used to replicate Kurath's study. Even though an item fulfilled one or more of the above conditions, it still may have been excluded. Sixty-one items, especially adjectives, that were included in the original plan proved in three trial interviews to be difficult to elicit without breaking up the flow of the questioning by focusing on words rather than content or by using an awkward "fill in the blank" type of strategy. Even without these items, Data 7 the interviews averaged three hours in length, and the volunteer informants often ran out of patience by the end of the sessions. Of the 179 remaining items, problems of comparability between data sets, incompleteness of the LAMSAS data, or complete lack of any variation in the responses necessitated discarding 29 additional questions. Though many of these had been included in the 1990 interviews, it became obvious after listing the responses that the two sets were not comparable, either because the questions as asked by Lowman and Johnson were different or because the fieldworkers perceived the target responses differently and thus counted different items as appropriate answers. For example...