1.6.2 The quality of post-edited texts

Jegyzet elhelyezéséhez, kérjük, lépj be.!

Another question concerning PE is the quality of post-edited texts. Quality can be studied in comparison to HT or independent of it. It is the former field that is more relevant from the perspective of the present research.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

As there is a widespread belief that it is easier to produce good texts from scratch than to edit and correct poor-quality texts, human translation is generally supposed to be of higher quality than post-edited MT. Existing research results do not support this hypothesis, as most studies have not found any difference between PE and HT (Daems et al., 2017; Garcia, 2010; Guerberof Arenas, 2008; Jia et al., 2019; Screen, 2016). Fiederer and O’Brien (2009) even found post-edited translations to score slightly higher on clarity and accuracy than from-scratch translations. Nevertheless, in the same study, HT was rated higher on style. In contrast, post-edited translations were found to be ranked higher in a study by Carl and his colleagues (2011) and to contain fewer errors than HT in the research carried out by Plitt and Masselot (2010). After NMT became widely available, the number of studies that showed the superior quality of PE texts over HT has started to grow (see, for example, Arjani & Jamshidiha, 2024; Lee, Sun-Woo & Sang-Bin Lee, 2021; Y. Yang et al., 2023).

Jegyzet elhelyezéséhez, kérjük, lépj be.!

The quality of post-edited texts, similarly to temporal efficiency, seems to depend on the domain of the source text. Lee and Choi (2023) compared the quality of HT and MTPE in the legal domain. Seven professional translators were involved in their research and were asked to translate a statute and to post-edit the DeepL output of statute translation. The direction of the translation was Korean to English (L1→ L2). The results showed that MTPE quality was better than HT quality. In addition, the comparison of the raw MT output and the post-edited MT outputs indicated that the translators could significantly improve the MT output (Lee & Choi, 2023).

Jegyzet elhelyezéséhez, kérjük, lépj be.!

In recent years, studies focusing on how different background factors influence post-editing quality have appeared, too. Zouhar and his colleagues have shown that better MT-output quality results in a lower number of post-editing changes (Zouhar et al., 2021), supporting evidence of the claim that MT-output is a major factor in influencing the post-editing process and product. Zou and his colleagues investigated how the translation brief (full vs light post-editing) and the search conditions (access to Termbase vs. access to the Internet) influenced post-editing performance (Zou et al., 2022). Although results depended on the competence level of the translators (experienced or not), full post-edited texts tended to have fewer errors, and primarily fewer fluency errors than light post-edited texts. Inexperienced translators made fewer accuracy errors when they had access to Termbase as compared to having the opportunity to search the Internet (Zou et al., 2022).

Jegyzet elhelyezéséhez, kérjük, lépj be.!

In summary, PE and MT seem to produce target texts of comparable quality. However, one shortcoming of these studies is a small sample size that generally leads to differences being small or invisible, in other words, statistically not significant.
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