This paper is essentially a consideration of methodological issues in L2 studies, with a focus on collecting production data in nonnative contexts. It begins with a discussion of past techniques for investigating L2 speech, and then makes a case for new ways of examining the speech of proficient speakers of second language varieties of English in an attempt to understand how they coconstruct English in multilingual settings. Drawing on a multiple case study that was conducted among a selected group of proficient speakers of English in Malaysia, the paper offers a critical description and evaluation of a technique used to gather spoken data generated in a variety of naturalistic contexts. In particular, it evaluates the election and preparation of participants, the research tools and procedures used, and the roles and relationships involved in the research. It is hoped that the issues raised in this paper will offer new methodological possibilities and serve as a resource for researchers in multilingual communities.
