1
. Speaking professionally in an L2

The fastest-growing use of globalised English is among speakers for whom it is not a first language, that is, English used as a lingua franca (ELF). To keep up with the developments of the language in such varying circumstances poses a challenge to research: how can we access reliable data that captures new directions in this expanding use of English? How should we go about securing enough data in a new area of language use, where variability is highly unpredictable, and change is likely to be fast? Clearly, corpus methods have a lot to offer in teasing out the big picture and emergent patterning from the bewildering detail that small-scale studies easily drown themselves in. ELF has established itself particularly in two important and influential inherently highly international domains: science and business. Both are high-stakes domains where language plays an important role. It makes sense to pay close attention to the ways English works in them and how it takes shape. This paper looks into the scientific sphere, and draws on the experience of compiling and analysis of the first ELF corpus, comprising academic speech (ELFA: www.eng.helsinki.fi/elfa). It will tackle issues of data selection, relevance, and meaningful combinations of analytical methods.