Across many disciplines, including sociology, anthropology, philosophy, cultural studies and sociolinguistics, writers and researchers are concerned with how language is used to construct representations of people in written and oral accounts. There is also increasing interest in cross-disciplinary approaches to language and representation in research. Within health, social care and housing research there is a rapidly growing volume of writing on, and sometimes with, people whose first language is not English. However, much empirical research in these fields remains at the level of 'findings' about groups of people with the issue of how they are represented remaining unexamined. In this article I discuss some of the different ways researchers have looked at issues of translation and representation across languages. As I show, some researchers have attempted to ignore or by-pass these issues in their research, some have given up the task as impossible and others have attempted the impossible. I argue that, although there can be no single 'correct' way for researchers to represent people who speak different languages, choices about how to do this have epistemological and ethical implications.