Book Review: Dimensions of Register Variation: A Cross-Linguistic Comparison

Many readers will be familiar with the quantitative and distributional techniques of register analysis presented by Biber (1988). The present book re-uses this earlier study on variation in British English [henceforth El, along with "a synthesis" (p. xv) of Ph.D. studies done with Biber at the University of Southern California: by Niko Besnier on Nukulaelae Tuvaluan IT], Mohamed Hared on Somali [S], and Kim Yong-Jin on Korean [K], plus other work, especially by Ed Finegan and Dwight Atkinson. What is new in this book is its cross-linguistic comparisons of register. Biber finds striking and "highly systematic similarities" in the patterns of linguistic variation across languages, and this leads him to propose "the possibility of cross-linguistic universals governing the patterns of discourse variation" (p. 359). The basis for this claim is the large differences between the four languages in their genetic affiliations, cultural histories, uses in literacy, and so on. T is Austronesian, with fewer than 350 speakers, on a small Pacific atoll. S is Afroasiatic, with 5 million speakers. And K is Altaic, with 65 million speakers. Appendices provide grammatical sketches of K and S, but not of T. And chapter 3 provides socio-cultural background on the four languages, some of it fairly rough and ready: for example, it is not mentioned that K is spoken in two different countries with very different political systems, and-I suppose different registers. And, as Biber admits, E consists of a highly variable set of geographical and social varieties. In chapters 1, 2, 4, and 5, Biber usefully overviews much previous work on register variation and the methods of register analysis that he developed in his 1988 work. He illustrates in detail the problems of trying to compare registers on the basis of isolated linguistic features. And he argues for an alternative multidimensional approach, where each dimension consists of a set of features that have been found to co-occur frequently in texts. He summarizes methods of grammatically tagging corpora, calculating the frequency of occurrence of features, using factor analysis to identify the dimensions, and then attributing functional interpretations to the dimensions. The main data comprise: 1. 8 present the main findings. First, Biber shows differences in register variation across the languages. For example: T has a restricted range of registers , all interpersonal in some sense, with only two written registers, personal letters and sermons, and no mass media; K and E have a …