What is striking about the changes in correlations between subcompetencies and performance across the two data collection waves is the growth in the number of significant correlations and the shifting importance of the subcompetencies. At the beginning of the training, language competencies seem to have priority; they show significant correlations with a number of performance indicators. A possible explanation could be that students have no other means to rely on, as they are untrained in translation. By the end of the second year of training, however, beliefs about translation dominate: they show the largest number of significant correlations with error numbers. In addition, thematic knowledge gains in importance and language competence (although still essential) recedes somewhat into the background. These changes between the first and the second data collection wave provide some tentative evidence on the dynamic restructuring of subcompetencies throughout the developmental process, as suggested by the PACTE group (2000, 2017). It is particularly important to highlight that correlations change in both the HT and the PE group, corroborating claims that traditional translation training will contribute to restructuring in post-editing competence, too. Clearly, as it was pointed out earlier, direct training in post-editing is definitely needed; nevertheless, devoting time to traditional translation training is not a waste of time either.