The present study forms part of the longitudinal study TransComp, a translation process study which investigates the development of translation competence in 12 students in the course of the Bachelor program in Transcultural Communication at the Karl-Franzens-University Graz. It aims at shedding light on reasons for errors which were made by students in translations at the beginning of their studies and repeated after three years of formal education in the field. The 12 students were divided into two groups of six and both groups had to translate 10 texts at different points of measurement. Thereby the very first text to be translated at the beginning of their first semester had to be translated once again at the very end of their sixth and last semester. Since the participants did not get any feedback on their translations it was expected that they might remember parts of the text but no solutions. The two experimental texts were popular science texts and had to be translated from English into German, the mother tongue of the participants. Data were collected using think-aloud, keystroke-logging, and screen-recording and transformed into translation process protocols (TPP) which also contain information regarding subjective problems (Krings 1986). The target texts were evaluated according to a primarily linguistic error classification scheme that differentiates between formal, semantic, grammatical, textual and other (culture-specific and idiomatic) mistakes.