The aim of this extremely thorough book is to provide both students and linguists with a comprehensive guide to the study of variation within language systems. This can often be a difficult area for students. However, James A. Walker combines theory and statistical analysis with naturalistic data to provide those interested with a one-stop-shop for the study of variationist analysis.
The scope of the book is wide-ranging and covers definitions of key concepts, analysis (including an entire chapter devoted to the GoldVarb statistical analysis tool), variation in sound and grammar systems, language change and language contact.
Walker initially sets out to provide a “storyline” for the teaching of the analysis of linguistic variation (p. xvii) and like any good storyteller, he starts at the very beginning. In Chapter 2, the author clearly defines the main terms used in the field of study. These include terms such as categorical and variable contexts, variable, variant, and the principle of accountability. Illuminating examples are provided from a range of languages and language varieties. Once the foundations have been laid, Walker proceeds to present fully exemplified step-by-step guides to the more common analytical tools used by linguists. In Chapter 3, he covers areas such as calculating frequencies and testing for statistical significance using chi-square tests (with the mathematics underpinning the statistics explained on a “need-to-know” basis).
Chapter 4, is devoted entirely to the use of the GoldVarb statistical programme. This could almost function as a stand-alone users’ manual to the programme. Examples and fragments of files are displayed precisely as they appear on the computer screen. However, the reader may be left with a number of questions insofar as it is sometimes unclear how certain codes appear in the coding string. When entering data into the GoldVarb programme, it is necessary to assign a code to each factor under investigation. If four factors are to be investigated, they could be coded as a, b, c, and d. The coding string (or token) forms the instructions for the statistical programme and is simply a conflation of the individual codes. Therefore the coding string for the aforementioned factors would be abcd. This is well illustrated by Walker; however, the example is less clear. In 4.1 (p. 32), the author presents a possible set of codes for t/d deletion. This is followed by figure 4.1 (p. 33) which shows a fragment of a token file for t/d deletion as it appears on the computer screen. New codes are in evidence here, and it is not obvious where they come from. Further explanations would have resolved the issue.
Having covered the fundamentals of establishing variables, variants, and factors, as well as having shown how to apply and interpret statistical analysis, over the next two chapters Walker leads the reader through variation within differing aspects of the linguistic repertoire. He reiterates that his approach to variation is a language-internal approach—that is, one which is not concerned with social or symbolic features of the utterances, but rather with language-internal factors such as the morphological status of the word in which the variant occurs. The first aspect approached, in Chapter 5, is the sound system which is also the most widely studied. In order to overcome difficulties in differentiating between variation in phonetic and phonological systems, Walker conflates the two terms to establish the umbrella term “phonic” variation. Using multilingual data, although heavily relying on t/d deletion in varieties of English, the author demonstrates, in a very practical manner, how to conduct research into phonic variation. Additionally, the reader is made fully aware of theoretical and methodological problems which may arise and, more importantly, how to overcome them. One area emphasized as potentially problematic is the researcher’s assumption regarding the underlying contexts of the variable. The example given (p. 47) is that of the plural form of the word ”ghost” in African Nova Scotian English. Generally speaking, it is realised as [gosəz]. At first glance, it may seem as if the final /t/ has been deleted; however, deeper analysis indicates the underlying form as [gos] for this particular variety of English. From...