This paper aims to contribute to the understanding of the implications of the use of audiovisual translation (AVT) as a language learning tool by analysing how the multiple layers of mediation inherent to an AVT learning activity can affect such activity, those who participate in it, and the final result. The article therefore presents qualitative data and results from a collaborative AVT project conducted by two Irish universities: NUI Galway and Dublin Institute of Technology and then proceeds with the analysis of the data, placing it within the framework of activity theory (Engeström, 2015). The analysis will illustrate the degree of complexity of a collaborative AVT learning activity by providing a graphical representation of the selected unit of analysis and then focus on the disturbances observed during the study, highlighting potential contradictions in the AVT learning activity. Finally, possible solutions and suggestions for further research will be presented.