The methodological task of defining 'translation' across languages forms an apparent aporia, since there is no guarantee that the different terms that might express 'translation' are in fact translations of each other. One solution is to propose a formal conceptualization of 'translation', in practice a set of criteria, a research filter, that the scholar imposes on the prior multilingual data. However, this act of imposition may neutralize the dynamic variability and historicity of the many culturally different ways of thinking about translation. Here the extent of that imposition is gauged by taking three examples of formal conceptualizations, from Toury, Gutt and Pym, and testing them on three cases of potentially borderline translational practices. It is found that the formal conceptualizations are not simple cultural impositions and can in fact allow considerable space for the historical study of textuality, receptive positions, and the semi-concealed subjectivities of translators.