Interestingly, the PACTE group (2017a, 2020) stressed the interconnected nature of its competence model, which implies that add-on skills are probably “embedded” items, and consequently, studying them together could bring more enlightening results than investigating them isolated from language competence.

When focusing on language competence, it is clear that reading and writing stand out as fundamental skills needed for translation and post-editing. Perhaps, the role of reading skills is even more important at the age of AI, when much of the process of target text creation consists of reading and comparing texts. In addition, reading plays a key role in acquiring thematic knowledge and in research and terminology work.

In recent years, some studies have included language competence as a variable in research on post-editing. Most of these studies, however, had language learners in their sample, as post-editing has become an activity often carried out by non-professionals. Below, studies incorporating language competence in some way will be reviewed.

Chung (2020) studied 59 Korean university students, who were assigned to three groups depending on their language proficiency level. Chung focused on the number and the nature of corrections introduced in the MT output during the post-editing process and found that high proficiency students made a significantly higher number of corrections than low English proficiency students. Moreover, the higher the proficiency level of the students was, the more likely they were to introduce changes above the word level. Low and intermediate proficiency students tended to accept the text-frame offered by the MT output and simply change words within the text. In contrast, high proficiency students often re-wrote phrases or even whole sentences (Chung, 2020). Chung’s final conclusions concern writing skills development and, as such, they are irrelevant from the perspective of translation competence. What makes it still important is that, exactly because of studying post-editing from a writing skill point of view, it clearly has a “post-editing as text-creation” approach, which is so much advocated by some translation scholars (Do Carmo & Moorkens, 2020; Rico Pérez, 2024; Robert et al., 2022). Ironically, the conclusion is that the higher the proficiency level of the student, the more likely they are to use MT output truly for text creation. Low proficiency students seem to have a revision view (keep the MT-text as a whole) and introduce fewer, and primarily lexical corrections, producing less acceptable texts (Chung, 2020).

In a similar investigation, Shin and Chon (2023) studied second language learners’ post-editing strategies when working in the direction of L1

search

L2. 57 English Language Teaching majors from two Korean universities formed the sample of the study, and they were categorised into B2 and C1 groups. The research findings suggest that students’ primary strategy is paraphrasing, often irrespective of the nature of the error in the text. Furthermore, the higher proficiency students were more successful in identifying errors in the MT output and also in using the appropriate strategies to correct them. Shin and Chon (2023) also reported instances of overcorrection, and they argued that limited language proficiency may explain this problem when translating from L1 to L2. They observed that students tend to overuse words, phrases and structures they feel safe with, but because of their limited language proficiency, this is a limited range. Students very often do not recognise correct but infrequent structures or uses of words and structures and substitute them with their safe choices (Shin & Chon, 2023).