In the first data collection wave, it was found that the more difficult HT students found the terminology of the task, the less accuracy errors they committed, and their total error number was lower, too. In the second data collection wave, it was the increased perceived thematic difficulty that was associated with a lower number of terminological errors. The only tentative conclusion we can draw from these changes is that the micro-processes of translation change as a function of growing competence, and the weight of the different sub-competencies changes.

In the PE group, the most striking finding is the increase in the number of significant correlations between PE-specific experiences and performance indicators. These correlations were analysed in detail in Chapter 3.4.4.2 Correlations between indices of task perception and performance measures. What is relevant for the developmental analysis is the growth in the number of correlations. A tentative explanation for this could be that students’ opinions and perceptions became more marked and stable as a result of training and experiences with translation, and this, in turn, leads to a larger number and more stable correlations. An intriguing phenomenon is that part of these correlations suggests that misconceptions and misrepresentations of the post-editing process may be formed as part of the developmental process. The correlation chain (Chapter 3.4.4.2) indicates that those who think that the task is easier and faster spend less time with the task, which is then associated with a higher number of terminology errors is a good example of that. Hopefully, this represents a transitional phase that will be overcome as soon as students gain more experience with post-editing.

Fortunately, some more positive patterns arise too. Apparently, there is a group of students who are sensitive to the creative nature of the translation task and mourn the loss of it. Increased sensitivity to the loss of the creative nature of translation was associated with higher overall performance and a lower number of fluency errors.