This volume is a collection of 10 chapters by leading specialists in applied linguistics. Generalizability is defined as “the extent to which research results can justifiably be applied to a situation beyond the research setting” (p. 3). Inferences are the links that the researcher makes between observations of the learners' data and their interpretation as evidence of theoretical arguments.

The first part opens with two chapters on testing. Deville and Chalhoub-Deville contrast the traditional way of conceptualizing variability with a more promising person- and context-sensitive alternative. In “Validity and Values: Inferences and Generalizability in Language Testing,” McNamara stresses the constructive nature of quantitative and qualitative research methods to focus on the distinction between the criterion domain (the target of test inferences) and the test itself (the evidence from which the inferences are drawn). The inferences derived are largely determined by the essential characteristics of the criterion domain. The key terms are then discussed in the context of theories of vocabulary acquisition (Chapelle). Test results inform us only indirectly about the learner's current mental lexicon. Therefore, the construct of vocabulary must be defined in the context of an adequate theory of vocabulary acquisition, which is still evolving. Duff then explores the operationalization of terms for qualitative versus quantitative research types, which are aptly systematized and contrasted. Swain writes on the interpretation of verbal protocols in SLA research, looking at the issue from the point of view of information processing and sociocultural theory, according to which verbalization can reshape cognition and experience. Larsen-Freeman addresses the theme of the volume in the context of functional grammar to stress the significance of complementary sources of data (e.g., contextual analysis) as helpful in understanding language acquisition and rationalizing language teaching. Markee provides the conversation-analytic perspective of the role of quantification and generalizability in SLA. His chapter contains a review of the literature on the interaction hypothesis, a critique of some experimental approaches to conversational analysis, and comments on function and organization of repair in language learning.

The issues presented in the seven chapters are next put into perspective. Bachman links them with a broad outline of generalizability in applied linguistics. Empirical research reflects a number of assumptions: It depends on the researcher involved (monism/dualism, emic/etic perspectives), the purpose (to describe the phenomenon, to induce some links with other phenomena, to explain it, to inform or falsify a theory), the intended audience (lay or professional), the constructs investigated (the trait, behaviorist and interactionist perspectives), their origin (from observation or theory), their level of specificity, the role of context in research, and the way of reporting (what counts as observation, how it is reported, and the unit of analysis), as well as the interpretation of findings (the operationalist, realist, or constructivist stance). Gass ponders on what it is it that we are generalizing in SLA, in which the plurality of approaches is desirable, or even necessary. She quotes data from her earlier study, which compared recalls of experienced and preservice teachers and points out that replication might be a way toward a deeper understanding of the phenomenon of learning. Hornberger looks at the critical points in the research on acquisition that demonstrate the insufficiency of the available research methodologies and conceptual tools. The purpose of research is to transform practices of language learning and teaching while maintaining humility and respect for the subject matter.

Indeed, emanating from this fascinating volume is an air of respect for the complexity of the subject matter—language learning and use—as well as the researchers' humility in its investigation. The volume might not be a paragon of editorial work (the order of the chapters seems rather random, with several misprints scattered throughout; one list of references would be preferable to individual lists after each chapter) but, meritwise, it certainly is a cutting-edge publication. It is symptomatic that it appears concurrently with another volume in the same series, edited by Norris and Ortega (2006), that contains meta-analyses of language learning and teaching studies. This volume addresses the awareness of SLA researchers of the promises and limitations as well as the contextual ramifications of their research methods, but first and foremost, of the constructive role of the researcher. Following nearly a decade of focus on theory construction, this bottom-up trend is a sure sign of the further maturation of language research. This volume is a highly recommended piece of reading.