Linguistic variation and language attitudes in Mannheim-Neckarau By Winifred V. Davies (review)

BOOK NOTICES 417 fects, and binding theory. The data show significant acceptability differences that accord with the assumptions in the syntactic literature but also reveal additional effects. C concludes that an experimental approach can uncover linguistic regularities that have gone unnoticed under the informal, intuition-based approach. The data show that the overall judgment patterns within a population are highly stable but also reveal considerable variation within individual speakers. Similar to other types of behavioral data, linguisticjudgments seem to exhibit a certain amount of random variation around a stable mean. This constitutes a potential problem for an informal approach to data collection, which does not provide adequate variance control, as standard methods of experimental design and statistical evaluation are not applied. In Ch. 6, 'Response methods and scaling issues' (67-77), C argues that acceptability is best measured using a ratio scale. Such a continuous, open-ended scale yields maximally delicate data and allows for the application of parametric statistical tests. Ratio data can be obtained using magnitude estimation (ME), a standard experimental paradigm in psychophysics for measuring subjects' perceptions of sensory stimuli (loudness, brightness, etc.). Subjects are required to make judgments proportional to perceived stimulus intensity, e.g. by assigning numbers or by drawing lines of appropriate length. ME yields robust and delicate results for linguistic stimuli, as shown by Ellen Gurman Bard, Dan Robertson, and Antonella Sorace ('Magnitude estimation of linguistic acceptability', Language 72.32-68, 1996). C proposes to use scannable line drawing (SLD) as a technique for collecting ME data. Instead of drawing an actual line, subjects mark the end of the line by ticking a response field on a standardized questionnaire. The questionnaire can be evaluated automatically, thus facilitating data collection. However , SLD only allows for a discrete number of line lengths (ten in Cs case) and hence fails to provide a continuous, open-ended scale. It limits the response ratio to 1 : 10 (if used optimally), whereas other studies found subjects to use considerably wider ranges, with 1 :9.9 as the mean ratio (Bard et al., op. cit.). Hence it seems likely that SLD seriously distorts the response scale, probably reducing it to an ordinal scale. C obtains a significant correlation between SLD and numeric ME (r2 = .85), but this is not sufficient to show that SLD yields ratio data and is able to make the same distinctions as standard ME. The remainder of the book is a tutorial on the design and statistical evaluation of acceptability judgment experiments, based on Cs methodological findings. The proposed approach can be realized efficiently and is accessible to linguists with limited background in experimental psychology. C deals with constructing sentence materials and questionnaires , issues of sample size, and presenting stimuli. The statistical part introduces data analysis using ttests , ANOVAs, and correlation measures. An appendix describes the use of spreadsheet software for constructing questionnaires and processing response data. Cs approach to linguistic data collection and his methodological findings are largely independent of his SLD technique and the potential problems with it. On the whole, C provides valuable guidelines for the experimentation with linguisticjudgments, along with an accessible introduction to relevant psychological methods. [Frank Keller, University ofEdinburgh .] Linguistic variation and language attitudes in Mannheim-Neckarau. By Winifred V. Davies. (Zeitschrift für Dialektologie und Linguistik Beiheft 91.) Stuttgart: Franz Steiner Verlag, 1995. Pp. xiii, 224. A revised version of the author's doctoral dissertation , this book grew out of sociolinguistic research carried out in southern Germany in conjunction with the Institut für deutsche Sprache in Mannheim. It follows in the footsteps of similar studies done in Berlin in the 1980s and attempts to investigate language attitudes, i.e. the social correlates to linguistic variation. The first chapter, 'Introduction' (1-2), briefly discusses the purpose of the work. The book really begins with the second chapter, 'Previous research into linguistic variation and attitudes' (3-32), which deals with a variety of historical questions. First, some of the principles of traditional dialectology are reviewed. Then, some of William Labov's sociolinguistic ideas are examined. Davies then discusses some Labovian studies that have been done in Germany and Austria and concludes by examining three earlier sociolinguistic studies done in Mannheim...