Márta Lesznyák: A longitudinal investigation of translation trainees’ translation and post-editing competence development I
›
3 Results and Discussion
›
3.2 The discussion of novices’ results
›
3.2.2 Performance indicators in the first year of training (time on task and error numbers) (research questions 1.4 and 1.5)
›
3.2.2.2 The quality of the target texts (error numbers)

3.2.2.2 The quality of the target texts (error numbers)

Jegyzet elhelyezéséhez, kérjük, lépj be.!

The quality of the target text is clearly the most important indicator of translator and post-editor performance. In this research, an MQM-based error typology was used to evaluate the target texts produced by the translation trainees. The error categories employed were accuracy, fluency, style and terminology (for more details, see Chapter 2.1.2 Materials).

Jegyzet elhelyezéséhez, kérjük, lépj be.!

Before interpreting our data, it should be stressed once again that students had no previous experience with translation or post-editing when they participated in the first data collection wave. However, as they had a BA degree in modern languages and/or a C1 level language exam in English, they can be regarded as proficient language learners/users. Consequently, their results may give us clues on how non-professional translators without a translation degree could perform in a similar setting.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

As there is no benchmark to compare students’ performance to, it is again impossible to tell objectively how well students did on the task. Nevertheless, mean total error numbers were above 20 for both the HT and the PE group (27.09 and 22.17, respectively), which suggests a relatively poor performance considering that the ST was 127 words long. The researcher, who acted as one of the raters, also had the subjective impression that both translators and post-editors struggled with the text. This is not surprising as they had no previous experience in translating.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

More important than the objective evaluation of novices’ performance is the comparison between human translators and post-editors. Statistical tests revealed significant differences between HT and PE students on accuracy errors (HT mean = 11.44, PE mean = 7.87) and the total number of errors (HT mean = 27.09, PE mean = 22.17). These figures suggest that translation students at the beginning of their training clearly benefit from post-editing: the overall quality of their post-edited target texts surpasses that of translations produced in the HT condition. In particular, post-edited texts are more accurate than from-scratch translations.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

A plausible explanation we can give for these results is that although NMT obviously cannot produce error-free translations, its quality is probably better than that of target texts created by translators without experience. Another possible explanation is that students in the PE condition have more capacity to check accuracy, as they have a first draft to work with. In contrast, in the HT condition, a lot of energy is spent on producing the exact first draft.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

The lack of significant differences between HT and PE students’ performance on other factors (style, fluency, terminology) may indicate students’ indifference or lack of awareness concerning these factors. It is exactly training that is expected to raise this kind of awareness and motivation.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

If we generalise these findings to non-professional translators (i.e., high-proficiency language learners and users), we may claim that MT and PE could be beneficial for them, provided they work with a text the topic of which they are familiar with. Post-editing may help to produce better and more accurate translations.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

The findings on the differences between PE and HT students support previous studies showing that PE texts are of better quality than HT texts (Arjani & Jamshidiha, 2024; Carl et al., 2011; Fiederer & O’Brien, 2009; Lee, Sun-Woo & Sang-Bin Lee, 2021; Plitt & Masselot, 2010).

Jegyzet elhelyezéséhez, kérjük, lépj be.!

In addition to comparing HT and PE students’ performance, the most frequent error categories were also identified in both groups. The results of the ANOVA tests showed that students’ target texts contained significantly more accuracy errors than any other types of errors, irrespective of the work mode they were engaged in. Unfortunately, no previous studies with the same design could be identified in the literature; as a result, it is hard to determine how common it is for novices to struggle with accuracy. Nevertheless, Quinci (2023), with a different design and a different quality assessment method, also found high means of accuracy errors in her novice sample.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

To sum up, quality assessment reveals that accuracy is a key factor in novices’ translation and post-editing performance. It is not only the most common error type, but also the factor that differentiates between post-editors and translators. It is also clear that inexperienced translators could benefit from using MT, provided they post-edit the text, as they would most probably prepare more accurate target texts.
chevron_left
chevron_right
Tartalomjegyzék navigate_next
Keresés a kiadványban navigate_next
A kereséshez, kérjük, lépj be!
Könyvjelzőim navigate_next
A könyvjelzők használatához
be kell jelentkezned.
Jegyzeteim navigate_next
Jegyzetek létrehozásához
be kell jelentkezned.
    Kiemeléseim navigate_next
    Mutasd a szövegben:
    Szűrés:
    Kiemelések létrehozásához
    MeRSZ+ előfizetés szükséges.
      Útmutató elindítása
      delete
      Kivonat
      fullscreenclose
      printsave