This study attempts to find an answer to whether normalisation is a feature of translation by using corpora-technology. Normalisation has existed as a proposed universal feature of translation for almost two decades, but so far, little research has been done on this issue. Some attempts have been made to provide evidence for (or against) the existence of such a universal, but these have failed in coming up with unquestionable evidence. The phenomenon of normalisation can be described as a tendency for translators to adhere to, and even exaggerate, typical features of the target language in their translations. In the present study, the verb take has been chosen as the object of scrutiny. A prototypical structure of the verb's semantic network has been developed, where the prototypical senses are regarded as typical features of the target language. Furthermore, sentence examples containing the verb take have been extracted from two kinds of corpora; corpora containing English translations and corpora containing non-translated English. During an analysis of these examples, the senses of the verb have been identified and registered into different sense categories. After this process, we are left will highly manageable figures which can be used to compare the distribution of the senses of the verb take in translated texts with the distribution of the senses of take in non-translated texts.