This paper discusses the issue of representativeness in the creation of general translation corpora. In the course of a in-depth examination of the stages involved in the selection of texts which adequately represent the target population, it demonstrates that prototype categories are better suited than the all-or-none classical ones to reconcile existing theoretical statements about what constitutes legitimate data in translation studies with the new methodology developed by the corpus-based approach.

No abstract

In the paper, I argue that human cognition is vital for language, the creation of human understanding of the world, scientific or otherwise, and in the interaction between these two. I consider the link between the world, language and the mind as it is played out in the task of determining what constitutes the object of study for the field of translation studies. The basic argument is that determining what is to be included in the 'translation' category is like all other category judgements, a highly cognitive endeavour. As such, the structures and processes involved in human cognition are vital elements of the process of matching word to 'thing', and of grouping kinds. In the paper, I outline the basic philosophy/cognitive science framework in which my investigations are couched and present the results of two empirical studies. The data from the first study indicates a pattern of prototype effects for the 'translation' category, which is subsequently supplemented with etymological data in order to derive a cognitive model for 'translate'. In the final discussion, I demonstrate the power of this cognitive model in addressing resilient (meta)theoretical dilemmas in the field of translation studies, including the position of interlingual translation as compared to other types of translation, the question of disciplinary boundaries and relationships, and the current position and potential utility of the equivalence concept.