Research with EEG and Eyetracking
We conducted several studies with the help of EEG and Eyetracking.
Reception of post-edited texts by students of different disciplines („Rezeption von post-editierten Texten – Studierende des Fachs Translation und Studierende technisch-orientierter Fächer im Vergleich”)
by Ilse Reusch
Post-edited texts are often considered as low-quality texts compared to human translation. In this study, the reception of post-edited texts were compared to manually translated texts via an online survey. The participants of this survey were partly translation students (n=14) and partly students of engineering-related disciplines (n=13). They had to rate text excerpts that were either post-edited or manually translated, but without knowing how the texts was translated, according to specific criteria like comprehensibility, style, general acceptability etc.
Are laypeople better post-editors? Post-Editing of Technical Documentation – A Case Study („Sind Laien die besseren Post-Editoren? Post-Editing in der Technischen Dokumentation”)
by Franziska Hör
This keylogging and screen-recording study compared the post-editing behaviour of six translation students with six translation laypeople. All participants had to post-edit one excerpt of a manual for a laser and fill out a short questionnaire in advance. The target texts were analyzed for mistakes (spelling, grammar, punctuation, content, terminology, style, and consistency), the keylogging data for editing effort and post-editing time, and the screen-recording for research behavior.
The influence of the source text on the post-editing efficiency – a qualitative study with eyetracker and Translog („Produktivität beim Post-Editing unter Einfluss des Ausgangstexts – Eine qualitative Studie mit Eye Tracker und Translog“)
by Karen Leicht
The aim of this eyetracking study was to assess how important the source text is to the post-editing task. In total, twelve translation students participated in this study. They had to post-edit four medical texts, which were presented in four different settings. In the first setting, source text and the machine translated (MT) output were presented on one screen for post-editing. This was also true for setting two and three. However, in these settings the participants were forced to read the source text (setting 2) or the MT output (setting 3) before they started to post-edit the MT output. In the last setting, the participants did not have the source text available at all and had to post-edit the MT output without the source text. Further, a retrospective interview was conducted. Post-editing time, keylogging data, research, eyetracking data (total fixation duration and fixation count), mistakes in the final target text, and the interviews were assessed in the analysis of the recorded data.
Cognate facilitation effect in bilingual and trilingual students - An ERP and RT study
by Eva Maria Flucke
It is a well established phenomenon that cognates (translation equivalents with a similar form in two or more languages – System (German) / system (English)) are translated faster due to their form overlap. This effect lead to the research question whether trilingual cognates are processed even faster than bilingual cognates.
For the empirical study, trilingual students (German/English/Italian vs. German/English/non-roman language) participated in a word translation test. They had to translate non-cognates, bilingual cognates (German/English) and trilingual cognates (German/English/Italian). Reaction times were measured and an EEG was recorded.