Tobii Case Studies
Lost in translation: Eye tracking in translation process research
Recently, translation process research has become a focal topic in translation studies. However, the investigation of how translations are perceived by the target audience remains still a research gap in process-oriented research. That’s where our studies come into play: we investigate whether the effect of translations is the same compared to originally produced text in the target language and if not, where the differences are. We triangulate the process data with comparable and parallel translation corpora. The latter provide correlations, cooccurrences and patterns, the first give insight into possible explanations.
Our research questions are the following: Is there a difference between translations and originally produced texts in the target language? Eye tracking helps us to investigate problems, peculiarities and untypical realizations in the translations. Indicators might be longer fixations, more fixations and regressions. The innovation of our work lies in the focus on the reception of translation. More concrete research questions can be formulated as follows:
- Are translations read in a different way compared to originals in the same language?
- How can we create efficient subtitles?
- How good is sign interpretation realized? Etc.
Moreover, we are interested in the translation and interpreting process itself, which leads us, for instance, to the following research questions:
- How is a source text read by a translator?
- How much focus is laid on the target text?
- Which visual input does an interpreter need?
Tools and Methods
Up to now we have carried out several studies in which we investigate the reception of translations by the target audience. We use eye tracking with Tobii TX300 Eye Trackeras well as retrospective interviews in order to triangulate data on conscious and unconscious processes involved in producing and perceiving translations. For the analysis of the data we use Tobii Studio Eye Tracking Software in order to calculate fixation duration, fixation count, time to first fixation, regression paths, etc.
Our results show several tendencies: Our first study compared the use of ordinary subtitles and integrated titles in American-English films. The results show that fixation duration, fixation counts as well as length of saccades are shorter for integrated titles (see figure below, right) compared to ordinary subtitles (see figure below, left). Integrated titles seem to be more efficient leaving more time for the actual film and protecting the aesthetics of the film (Fox 2012).
In a second study we investigated the perception of sign interpreting for Phoenix news on TV. The results indicate that the there is an overload of information for hearing-impaired people. Furthermore, the presentation of the sign interpreter is insufficient since it appears to be too small and not centered. The newscaster who actually reads aloud the news is not important within this context and should be replaced by the sign interpreter (Gutermuth 2011).
Finally, first results in translation process research corroborate the assumption that reading for translation differs from other reading tasks. Translators’ reading behavior (see figure below, right) is not as linear as regular reading (see figure below, left). The regressions while reading in a translation tasks show that translators analyze compound nouns and named entities or abbreviations while reading and they jump back in order to figure out the cohesive ties of the source text (Hansen-Schirra 2013).
Further studies comprise the investigation of:
- linguistic Web usability
- the usability of translation technologies
- the comprehensibility of expert-lay communication (popular-scientific texts, instructions, legal texts)
- non-native reading
- the efficiency of machine translation postediting for professional translation
- the integration of text and images
"With respect to the heterogeneity of our projects and huge amount of data involved the Tobii Studio Eye Tracking Software reaches its limits. In future, we will therefore rather process the raw eye-tracking data", explains Prof. Dr. Silvia Hansen-Schirra. "However, since we use eye tracking not only for research but also for teaching purposes and for MA and BA theses our experiences with setting up and carrying out an experiment with Tobii Studio are really good. After a brief introduction, BA and MA students are able to work with the eye tracking software and hardware enabling them to carry out own research projects. Moreover, we can use the Tobii TX300 Eye Tracker in very flexible settings, e.g. as standalone unit in an interpreting cabin."