The notion of some sort of translational 'change' has long been a central issue in Translation Studies. Some studies (e.g. Catford 1965, van Leuven-Zwart 1989, 1990) have resulted in detailed frameworks that have subsequently been used in empirical studies. Similarly, Vinay and Darbelnet’s 'methodology' for translation (1958/1995) represents a detailed account of translational relationships that has been described as a taxonomy of shift types (Munday 2001:55ff). From a more current perspective, it may be argued that the ongoing research interest in translation universals, at least in some of its manifestations, also represents the same concern with ways in which translations differ from their source texts. In this paper, the author suggests an approach to translation shifts which may capture many of the insights of previous work, but which holds more explanatory potential. She approaches the question of translation shifts from the perspective of cognitive linguistics and suggest that translation shifts derive from so-called construal operations, and as such are fundamentally cognitive. The paper proposes a link between the various kinds of shifts posited in the Translation Studies literature and a construal operation that could ultimately explain them. The various types are illustrated with examples taken from the literature and from the Oslo Multilingual Corpus.