Insights derived from language corpora have influenced countless aspects of language teaching, testing, research, and materials development. Granger, Hung, and Petch-Tyson have created an indispensable testament to the specific value of corpora in SLA research. A great virtue of this volume is that the corpus is treated as a tool, and the tool does not distract from the job at hand. The focus is where it should be—on features of native language, interlanguage, and target language as elucidated by corpus-based study. The corpora stand in the background, ready to serve.

Granger's introductory chapter offers an overview of research done with learner corpora. Even corpus-savvy readers should take the time to read this chapter, which goes beyond a survey of corpus-design principles to a typology of learner corpora and a glimpse of the specialized software needed to work with them. This information proves helpful in reading later chapters. The other chapter that dedicates considerable space to corpora themselves is Meunier's discussion of pedagogy (the fifth of these unnumbered chapters). According to Meunier, teachers can find very different language patterns in native language corpora and learner corpora, such as learners' propensity to complement the verb accept with a full infinitive (to + V inf), a pattern absent in the native language corpus. Moreover, Meunier suggests that such discoveries should influence pedagogical decisions.

The other seven chapters highlight relatively narrow language phenomena. The distinguishing feature of these treatments is that each pattern under discussion is solidly evidenced in at least one corpus. Altenberg uses data from the English–Swedish Parallel Corpus to explore an apparent tendency of Swedish speakers to overuse the causative make when they write in English. Aijmer follows with a look at the use of English modals in the interlanguage of native speakers of Swedish, her evidence drawn from the component of the International Corpus of Learner English provided by Swedish learners of English.

The order of acquisition of English verb forms is reexamined by Housen on the basis of data from the Corpus of Young Learner Interlanguage. Housen depends on code-heavy extracts from the data, but the reader's patience will be rewarded. His conclusions, though not unexpected, suggest new possibilities for inquiry along a classic line. Later in the book, Hasselgren presents a very interesting study of the effect of smallwords (such placeholders as okay, and stuff, and delexical like) on the perceived fluency in English of Norwegian teenagers. Her data were drawn from the University of Bergen's EVA Corpus with the express purpose of improving judgments of fluency during language testing. I greatly enjoyed this chapter, even as I marveled at any positive correlation between fluency and a teenager's use of like.

Connor, Precht, and Upton contribute a well-argued discussion of genre moves as revealed by the Indiana Business Learner Corpus. As the authors point out, computers do not do a very good job of sorting data according to rhetorical moves. Humans have to tag instances by hand. This short chapter cannot open a large window on this process or on its dependability, but it suggests practical benefits for the learner of business English if dependability is achieved. Allan discusses ways in which learner corpora can help prepare future language teachers for the features of their students' writing (and, to a lesser extent, speaking). His data source is the TELEC Secondary Learner Corpus in Hong Kong. Seidlhofer, in the anthology's final chapter, describes what is essentially a learner mini-corpus composed of her Austrian students' writing in English. This was used as means of teaching her students how to mine corpus data in ways that might be valuable to them as teachers or researchers later in their careers.

This valuable anthology avoids the risk of being outmoded that awaits any book about corpus-related language research. The language points are solid in their own right and will remain so even after the software and databases used in their investigation start to seem quaint. Granger's is a major voice in corpus-related SLA research. The authors have compiled a great collection.