In lieu of an abstract, here is a brief excerpt of the content:

3 Change This study is unusual among sociolinguistic analyses because of its time span. Few variationist research projects have had truly comparable data spanning nearly sixty years available for analysis. Real time studies have an advantage over studies using change in apparent time (across age groups), since they do not have to rely on other sources of evidence to determine whether differences among age groups correlate with actual change in the language over time. Studies with teenage informants, in particular, run the risk that differential usage is a function of peer group identity rather than innovation that will have an impact on the language as a whole (Christian, Wolfram, and Dube 1988:139). This project will examine aspects of diachronic change that can be verified statistically. It will also investigate the relationship ofchange to variation. Change in the lexicon could be statistically documented for 232 variants that were used more frequently in one or the other of the data sets. Tests were conducted on each of the 670 variants that occurred at least three times in one of the samples. One hundred forty-five words were used significantly more often in the earlier data collected by Lowman; 87 were more common in Johnson's interviews. The variants that increased or decreased in usage during the century are listed in Appendix 3. [The "x" with some entries indicates that various contexts were combined. For example, "(x) bad weather" for the question 'clouding up' represents all of the following responses: fixing to be bad weather, going to be bad weather, looks like bad weather.] Twenty-two percent of the 670 variants subjected to statistical testing proved to be old; 13 percent of the words are new. In terms of which linguistic variables proved productive, 111 of the 150 interview questions (74 percent) exhibited at least one variant that was old and 67 (45 percent) had at least one new word as a response. Only 29 questions 62 Change exhibited no change in their sets of lexical variants, while 33 had among their responses both old and new terms. Speakers sometimes commented on words that they considered to be either modem or, especially, old-fashioned. Ofthe 670 variants that were tested with statistical methods, 113 (17 percent) were noted as old by at least one informant; 43 of these were documented as old by the analysis. Seventy-three variants were considered to be new by at least one speaker; only 16 of these were verified as new by the analysis. Thus, although speakers perceived many changes that were not documented, they do provide corroborating evidence for 30 percent of the old words and 18 percent of the new ones. It seems that speakers have more subtle knowledge about language change than could be verified quantitatively. Allen (1989) discusses such commentary at length. This chapter will examine those vocabulary changes that have been documented diachronically. The first section covers statistical methods used to verify the changes. The 67 variants that changed and that were also significantly linked to one of the demographic variables in the earlier analysis will be discussed, and the overall relationship between change and variation assessed. Finally, theories of lexical and semantic change and their applicability to the results of this study will be examined . Meillet (1921) was one of the first historical linguists to study the role of sociological factors in linguistic change. His work, as interpreted in Nerlich (1992:177), explained semantic change based on (1) relationships between words and other words, (2) relationships between words and objects, and (3) relationships between words and the differential usage of them by speakers of different social groups. The dimension of change that is linked to social groups is discussed first in this chapter, followed by the topic of change based on relationships between words. The relationship between words and objects is a major focus of Chapter 4, which deals with the influence of cultural change. Statistical Methods The binomial principle continues to apply in this part of the study (Schneider and Kretzschmar 1989). This requires that each variant or member of a set of responses be tested separately, as opposed to a multinomial test of the whole set at once. The database columns that were filled in with Is and 9s for the first analysis were used again in this part of the project. A table was projected from the responses to a question from the first sample with only Change 63 the rows where the relevant variant occurred (1) or...

Share