Márta Lesznyák: A longitudinal investigation of translation trainees’ translation and post-editing competence development I
›
3 Results and Discussion
›
3.1 Novices (1st year students at the beginning of their studies)
›
3.1.5 Translation/post-editing performance

3.1.5 Translation/post-editing performance

Jegyzet elhelyezéséhez, kérjük, lépj be.!

As described above (2.1.2 Materials), all target texts were evaluated by two raters and mean values were calculated from the two raters’ error numbers. All further analysis was performed with these mean error numbers.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

Table 12 presents the summary statistics for error types by work mode (HT or PE), and it also compares the two groups with the help of independent samples t-tests. As can be seen from the table, only the accuracy error numbers and the total number of errors differ significantly in the two groups (p < 0.05). It is also important to note that practically, differences in total error numbers can be accounted for by accuracy errors. In other words, post-editors’ target texts are of better quality overall, and this is because there are fewer accuracy errors in their texts.
 

Jegyzet elhelyezéséhez, kérjük, lépj be.!

Table 12 Differences Between Human Translator and Post-Editor Performance in the First Year. Results of the Independent Samples T-Tests
 
Human translators (n = 16)
Post-editors (n = 15)
 
 
 
M
SD
M
SD
t-test
p
Accuracy errors*
11.44
4.30
7.87
3.85
2.42
.022
Fluency errors
5.41
2.68
4.83
1.95
0.68
.504
Style errors
6.63
2.673
6.03
2.17
0.67
.506
Terminology errors
3.63
1.24
3.43
0.92
0.8
.632
Total number of errors*
27.09
5.15
22.17
4.53
2.82
.009
Note. M = mean, SD = standard deviation, * factors with significant differences
 

Jegyzet elhelyezéséhez, kérjük, lépj be.!

Another important question is whether certain types of errors are more prevalent than others within a sub-sample. Repeated measures one-way ANOVA was carried out to find out whether differences between error numbers within a subgroup were significant.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

In the human translator group, the Mauchly’s test indicated that the assumption of sphericity had been violated (χ2(5) = 15.857, p < .01), therefore, the repeated measures one-way ANOVA was performed with a Greenhouse-Geisser correction, which determined that some mean error numbers differed statistically significantly (F(1.926, 28.895) = 19.382, p < 0.0005). Post hoc analysis with a Bonferroni adjustment revealed that accuracy errors were statistically significantly higher in number than all other types of errors. In addition, there was a significant difference between the number of stylistic errors and terminology errors. For the sake of readability, results of the pairwise comparisons are reported in detail in Appendix 9. With respect to terminology errors, it should be noted again that terminology errors were counted just once, even if there were multiple occurrences of the same, misused term. This may have contributed to unrealistically low terminology error numbers in the study.

Jegyzet elhelyezéséhez, kérjük, lépj be.!

In the post-editor group, data violated the assumption of sphericity, too (Mauchly’s test: χ2(5) = 16.704, p < 0.01), as a result, degrees of freedom were corrected using the Greenhouse-Geisser estimates of sphericity. Repeated measures one-way ANOVA showed that there was a significant difference between error numbers (F(1.617, 22.636) = 8.336, p < 0.0005). Pairwise comparisons with Bonferroni adjustment indicated that there were significantly more accuracy errors than terminology errors and fluency errors (although the latter finding is only marginally significant). Just like in the human translation group, significant differences were found between stylistic errors and terminology errors. Results of the pairwise comparisons are reported in detail in Appendix 9 First data collection wave (first-year students).

Jegyzet elhelyezéséhez, kérjük, lépj be.!

The results of the ANOVA tests in the human translator and the post-editor sample suggest that accuracy errors stand out as dominant error types in both groups, although it is less striking in the post-editor group. Differences between other error types are not significant or can be measurement artefacts (i.e., terminology errors).
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