The application of eye-tracking in translator training

By Michał Kornacki (University of Łódź, Poland)


The act of human translation is a process in which source data is analysed, processed and the target data produced. The act itself, although similar, is unique in details for every translator. The concept to study this process is not new. In fact, most contemporary studies in the field of Translation Studies are examples of process research (see, for instance, Göpferich 2009, Saldanha and O’Brien 2013). The following paper focuses on one type of process research methods, i.e. eye-tracking, in order to determine its applicability in a computer-based translation classroom. Technology is an irrepracable element of translator’s workshop and it is critical to see how it affects professionals. Computer-assisted translation tools have moved from MS Word plugins to fully fledged GUI based standalone computer programmes. As a result, their visual aspect that makes them an ideal environment for a research based on eye-tracking methodology. Finally, the paper attempts to define new challenges for transator trainers and how they affect both course content and approach to new technologies in the classroom.

Keywords: translator training, eye-tracking, computer-assisted translation

©inTRAlinea & Michał Kornacki (2019).
"The application of eye-tracking in translator training"
inTRAlinea Special Issue: New Insights into Translator Training
Edited by: Paulina Pietrzak
This article can be freely reproduced under Creative Commons License.
Stable URL:

The face is a picture of the mind as the eyes are its interpreter
Cicero, 1st century BC


1. Introduction

The act of human translation[1] is a process in which source data is analysed, processed and the target data produced. The process varies in details for every translator, even though its general principles are the same. Each person has a different background, personal and professional experience, work code and preferred tools. As a result, both the approach to the source document and individual performance vary for each translator. Such statement leads to another reflection – there is always something new to learn and, therefore, translators are learners. As a result, it can be concluded that systematic application of process research in teaching may help to improve both process performance and learning strategies (Dam-Jensen and Heine, 2009) of future translators, which can be then re-used in professional life.

In fact, most contemporary research in the field of Translation Studies are examples of process research (see, for instance, Göpferich 2009, Saldanha and O’Brien 2013). The thing that changes is the approach to such research. Academics see the need to include authentic data coming from the actual practitioners of the trade, set in the context of rapidly expanding translator’s workshop and the redefinition of the profession of the translator. The problem they face lies in the fact that the majority of readily available data capture methods show certain trends regarding the profession, but lack more in-depth insights into minute details of the process, like attention paid to details, time frames and points of interest during the translation process. Methods like retrospective interviews, for instance, provide valuable feedback but are still outclassed by, for example, screen recording combined with comparative analysis (see Kornacki 2018).

Therefore, the first question that needs to be answered is: what methods will give us access to one’s mind? The obvious answer would be neuroimaging,[2] but it is too costly to be regularly applied in translation studies. The answer to the question can be found in words by Cicero, which open this very paper: ‘The face is a picture of the mind as the eye are its interpreter.’ Cicero notes that eyes reflect our mind and that we, humans, rely mostly on our sense of sight to take us through our lives, especially so today with the ever-increasing volume of visual data. And while it could be argued that ‘our body’ talks, the only part of our body that can provide information regarding our cognitive processing are our eyes. As a result, a research platform is required to allow academics to record the processes arising from our mind, ones that can be registered through our own eyes.

The following paper takes a closer look at one of the process-research methods – eye-tracking – in order to determine its applicability in a translation. The paper starts with an outline of the eye-tracking environment and application of eye-trackers in translation in an attempt to turn reader’s attention to the problems a translation trainer might face. Next, the paper proposes that eye-tracking can be used to greatly improve self-regulatory skills of graduate translators through exposure to actual issues, resulting both from human actions and machine operation. The paper closes with a brief discussion of new challenges for translation trainers in the nearest future.

2. Application of eye-tracking in translation

Introspection (think-aloud protocols, retrospective and introspective interviews), keystroke logging, screen recording, eye-tracking, personality profiling, contextual inquiry, and physiological measurements are some of the process research methods utilised in Translation Studies. Interestingly, even though the methods mentioned above are used primarily in academic research, some of them may be quite successful when used for translation teaching. However, their applicability as individual procedures is limited (Saldanha and O’Brien, 2013). The one method that probably deserves more attention is eye-tracking, which can be seen as an ultimate process research method because it combines keystroke logging and screen activity recording, adding gaze data capture on top of that. The purpose of an eye-tracker (a computer monitor-based device that records eye movement during a given activity) is to give researchers insight into how various eye muscles contract, thus allowing to draw conclusions regarding mental processes taking place in test subject’s mind. The phenomenon was widely exploited by psychologists up to date (see, for example, Wells 1792, and Hering 1879, as cited in Wade, 2015).

Modern eye-tracking derives from eye-mind hypothesis originally developed by Just and Carpenter (1980) (see also Holmqvist et al. 2011, Saldanha and O’Brien 2013, O’Brien 2015). The theory proposes that eyes seeing the data and the brain processing it are two concurrent activities. The data can be then used to assess cognitive effort, focus of attention (eye-fixations), and to use it to visualise gaze paths. As a result, ‘eye-tracking can be found useful in various domains, e.g. accessibility, usability, readability, psycholinguistics, information processing, emotion research, language learning, video-gaming industry, process research in translation’ (see Pietrzak and Kornacki, forthcoming) and translator training, as proposed in this paper.

2.1. Eye-trackers: an overview

Contemporary eye-trackers may resemble standard computer monitors, monitor-mounted devices (Tobii Pro X3-120, for instance) and even glasses (which are frequently used in research requiring head movement, e.g. during shopping). The technology used in most eye-trackers utilises ‘built-in infrared light diodes, which reflect light off the eyes, enabling the eye-tracking software package to calculate the precise X and Y coordinates of the eyes’ (Saldanha and O’Brien 2013: 136) on the eye-tracking unit. ‘The dilation (widening) or constriction (narrowing) of the pupil is also measured in pupil diameter per millisecond as are the rapid eye movements from one point to another (known as ‘saccades’)’ (ibid.). The ‘saccadic’ exploration of images, stationary or when scanning pictures, was pioneered by Alfred Yarbus (see Wade 2015). Contemporary eye-trackers offer sampling rates between ‘25 and 2000 measurements per second, which means that the faster devices achieve sub-millisecond temporal resolution, similar to EEG’ (Eckstein et al. 2017: 70) (which can be used as a form of physiological measurement, mentioned later in the paper). If we consider the fact that contemporary eye-trackers take the form of computer monitors or head-mounted devices, it becomes apparent that they surpass other process research methods since they can be used in a more natural environment (for the test-subject), e.g. schools or hospitals. The result is that it is possible to test more subjects and obtain high-quality data in short time. However, there are numerous issues a successful eye-tracking researcher/translation trainer[3] has to overcome in order to obtain the desired data (see, e.g., O’Brien 2009, Harezlak et al. 2014, Hvelplund 2014 for more details on handling eye-trackers and setting eye-tracking environment).

2.2. Application of eye-tracking in the translation environment.

Translation researchers agree that the development of eye-trackers that yield high-quality data has changed general approach to process research in Translation Studies. Hvelplund (2014: 203) notes that the technology has been successfully employed to conduct research on such topics as:

  • translation memory tools and cognitive load (O’Brien 2006);
  • reading for translation as a particular type of reading (Jakobsen and Jensen 2008);
  • coordination of comprehension and production processes in translation (Dragsted and Hansen 2008);
  • directionality in translation (Pavlović and Jensen 2009);
  • reading modalities in translation (Alves et al. 2011);
  • distribution of cognitive effort during translation (Hvelplund 2011);
  • translator competence (Ehrensberger-Dow and Massey 2013);
  • metaphor translation (Sjørup 2013);
  • classification of translator styles (Dragsted and Carl 2013);
  • and parallel processing in translation (Balling et al. 2014).

Most of the above-mentioned research would not have been possible had it not been for O’Brien (2009) and Alves et al. (2009), who have discussed problems related to methodology of eye-tracking-based research. Specifically, O’Brien discusses challenges encountered during the data-collection process (including, e.g., research environment) while Alves et al. deal with the issue of combining various process research methods with eye-tracking, as well as reliability of collected data (Hvelplund 2014).

It has to be noted that due to the nature of eye-trackers and amount of control a researcher has to exert over the process, the data capture is usually conducted over individual sessions. Translation seems to be a natural environment for eye-tracking research since translators use computers to translate every day. Therefore, replacing regular monitor with an eye-tracker would not be a significant change in terms of work conduct. At the same time, the fact that contemporary translators rely mostly on digital resources should not be underestimated. The computer screen contains all the data used during the translation process – word-processor/CAT tool, online dictionaries, web browsers, online resources, machine translation, etc. As a result, the translator does not have to refer to resources outside the computer screen[4] and can hold gaze on the monitor. Therefore, the entire translation process can be captured as screen recording overlaid with eye-tracking data and analysed as per research problem. By extension, the method can be applied to translator training, both in the form of data collected from professional translators (reference, benchmark) and student translators (student-level process capture).

The method offers a very detailed insight into the process of translation, including reading the source text, mining the Internet for terminology and context, interacting with various computer tools (Teixeira 2014), and finally rendering the target text.

2.3. Eye-tracking data

Translation Studies is only one of many fields that benefits from the eye-tracking research. Therefore, the present paper will focus exclusively on its application in the translation workshop. Below, the author presents types of eye-tracking data that allow to draw conclusions regarding the nature and process of translation.

The first type of data is the number of fixations, understood here as periods of eyeball stability, i.e. the eye is fixed on a given point on the screen. Duchowski (2007) sees their purpose as means to bring the observed object into visual focus. Hvelplund (2014) reports that fixation duration and fixation count are prevalent measures in eye-tracking-based translation process research and ‘they are often taken to index cognitive effort’ (ibid.: 212). Basically speaking, the longer fixations, the more effortful processing (more difficult task). By contrast, the shorter fixations, the less processing is required (an easier task). If combined with the fact that the target text (or at least the part of the computer screen displaying target text) is reported to attract more fixations than the source text area of the screen (see, for example, Jakobsen and Jensen, 2008; Pavlović and Jensen, 2009), we may get the impression that target text production is more difficult than source text reading and comprehension. Hvelplund (2014: 213) wonders if ‘these more and longer fixations over the TT [target text] area have to do with the eye moving more slowly across the TT in sync with the emerging TT being typed?’ He proposes the idea that it boils down to the fact that target text reading speed is directly related to the physical aspect of translating, i.e. typing speed (ibid.).[5] Therefore, the activities listed above cannot be directly correlated but instead should be considered different tasks measured with an eye-tracker. Such consideration may lead to some interesting research like, for example, comparison of fixations on target text in regular translation (using a word-processor-like MS Word), desktop CAT tool (e.g. SDL Trados), and cloud-based CAT tool (e.g. Memsource). When discussing fixations it is necessary to mention saccades, microsaccades, refixations and regressions. The notion of saccades refers to voluntary movement of both eyes between two or more phases of fixation in the same direction. This is related to the fact that fovea area of the retina is relatively small and saccades are used to maintain high-resolution view of the observed content. Saccadic eye movement is natural and should be considered as such when analysing fixation data. Unlike saccades, microsaccades are involuntry, jerky eye movements that ‘are the largest and fastest of the fixational eye movements. They contribute to maintaining visibility during fixation by shifting the retinal image in a fashion that overcomes adaptation, thus generating neural responses to stationary stimuli in visual neurons. […] Recent discoveries have shown that microsaccades are critically related to many aspects of visual perception, attention and cognition’ (Martinez-Conde et al. 2009: 463). Similarly to saccades, microsaccadic eye movement should be taken into account when analysing eye-tracking data. The last two types of fixation-related data are refixation and regression. Both concern backward gaze movement in the sense that eyes refixate on content that has already been seen. This may be result from corrective saccadic movement (e.g. correction of ‘overshooting,’ or problems processing the current word) or problems with text comprehension. As was mentioned before, all of those aspects of fixation data should be taken into consideration both during planning eye-tracking session and subsequent data analysis.

The second type of eye-tracking data is related to the size of the pupil. Holmqvist et al. (2011) suggest that changes in pupil size may indicate strain on the cognitive system. For example, more difficult tasks result in dilated pupils whereas in less difficult tasks the dilation is considerably lower. However, pupil dilation may yield false data and should be collected and analysed with caution. As was mentioned before in Section 2.1 (see also O’Brien, 2009), pupils change size due to many factors, most of them not related to cognitive stress. Light, emotions, medicines, stimulants (drugs, alcohol, energy drinks), illness, all affect pupil behaviour. Therefore, effort should be made to sort test subjects and exclude those who display symptoms of illness, consumption of stimulants, or unfit according to previously outlined requirements. Another issue to be taken into consideration pupillary response, or rather its aspect called pupillary latency. It is the delay between the visual stimulus and pupillary response, e.g. pupil reacts to light change between 150-400 ms after the change takes place (Holmqvist et al., 2011). Hvelplund (2011) notes delay of 120 ms for source and target text reading during the process of translation. The delay has to be considered when preparing eye-tracking research and reviewing research aims lest it results in faulty data. Hvelplund (2014: 215) suggests that

[t]his could be done either by applying a fixed pupillary delay to all recordings across all participants or by applying individual pupillary delays for each participant, assuming that not all participants’ pupils respond with the same delay. The latter approach requires that individual baseline measurements have been recorded for each participant before collecting the actual process data.

To make things harder, researchers have to take one more issue into consideration, hinted by Hvelplund above – the delay is different in different people and increases with age (see Feinberg and Podolak, 1965). In the end, all those issues have to be taken into account or the obtained data will have little relation to the base material used to capture the eye-tracking data.

The last two forms of data collection include visualisation of eye-movement (heat maps[6] and gaze plots[7]) and capturing of the total gaze times. Total gaze time gives us insight into the areas of the monitor that attracted the translator’s attention most often (word, phrases, panels, windows, etc.), thus allowing us to speculate on the process of translation, background and actions of the translator. Visualisation offers an accurate visual impression of which parts of text and screen received the highest number of fixations during the process of translation, unlike gaze times which produce hard numerical data. The problem with this form of data is that it cannot be tested, analysed and verified through statistics and, therefore, it is used primarily to generate hypotheses.

The above overview is to meant to show the complexity of the eye-tracking data and suggest that only a translation researcher can utilise eye-trackers for teaching purposes. The data is simply too complex and too difficult to collect to risk inexpert approach to the problem. Therefore, for the purpose of this paper, translation teacher and translation researcher are considered one and the same.

The following section proposes how to apply eye-tracking in a classroom. However, unlike previously outlined approaches, which were research-oriented, the next section advocates utilising the eye-tracking technology as a teaching aid.

3. Eye-tracking as a teaching aid in a computer-assisted translation classroom

Previous sections provided background for eye-tracking research and shed some light on the types of data we can collect and show to our students. This section takes on the topic of using eye-trackers as teaching aids in a computer translation classroom and tries to identify teaching problem areas that were otherwise neglected with no access to oculography.

The process of eye-tracking is time-consuming as specific preparations (hinted at previously) have to be made, and then the process itself needs to take place. For that reason, it would be difficult to run it for each student member of a group with respect to a single translation task. If it were to be the case, the teacher would have to arrange several recording sessions which may not be possible considering time constraints on behalf of students (it would have to take place in their free time[8]) and the teacher. If, on the other hand, the data capture was extended in time to accommodate time issues, the teacher could risk a situation in which the data captured at the beginning to be obsolete since students may have already made progress. A solution to the problem is to treat each student individually and capture and discuss the data with the students after it was captured. However, it may not be attainable due to time-constraints mentioned before. Therefore, the author sees it as a way to use individual examples to show some patterns and tendencies in the process of translation to a group of students.

Before any data is recorded and analysed, it is vital to introduce students to the concept of eye-tracking and show them sample data in order to familiarise them with the obtained output. Only then will students be able to draw conclusions and learn from the data presented to them. Such introduction can take the form of a general presentation outlining the process, types of data (e.g. fixations, heat maps) and its analysis. What is more, students should also have access to the recording so that they can reflect on the process by themselves.

In the classroom, eye-tracking can be used in two ways: to visualise specific patterns in the process of translation on examples of E-T data taken from individual students and to compare students to professional translators. However, the author of the paper sees the greatest value of the method not in showing specific cognitive processes taking place (although the data may prove to be interesting) but rather in visualising certain translation patterns in the process of translation. Therefore, the paper focuses on CAT environment which is a unique blend of translation (source and target texts divided into sentence-based translation units and juxtaposed one next to another) and assisting software (auto-translation, auto-glossaries, machine translation). The author believes that it is this area that justifies the use of such expensive equipment as eye-trackers in the classroom.

3.1. Discussion on samples of data taken from students

Group discussion on data taken from students has both advantages and disadvantages. The former include time-efficiency in capturing and discussing the data, ability to highlight specific patterns and tendencies that translator trainer sees in most of the students and wants to show to them, and the ability to rerun the process later in the semester to see if the tendencies remain as strong as before. The latter concerns the lack of student-specific data and inability to provide them with detailed comparative analysis of their results.

It is vital, however, to note the types of data of particular interest during such exercises. Changes in pupil size may not be as statistically significant here as the number and duration of fixations (with attention paid to refixations and regressions[9]) and, above all, visualisations (heat maps and gaze plots). In 2015-2017, the author of the paper run research on student’s over-reliance on computer-assisted translation (CAT) tools (see Kornacki, 2018) utilising screen recording. In the study, students were asked to record their translation process using BB FlashBack Express software. Two groups of 16 students were tested, the 2015/2016 group being a control group, and the 2016/2017 group being the test group. Both groups were to translate fragments of a financial statement (Polish to English, around 8,000 characters with spaces) using CAT tools (memoQ). The research was conducted in two parts, 1st after exposure to CAT tools, and 2nd three months later, after gaining some experience by students. In both cases students used Client-provided translation memory (TM). One of the goals of the research was to prove that students (and graduate translators, by extension) tend to over-rely on Client-provided content (TMs) to the extent that they fail to detect and correct TM-based translation errors, thus resulting in a faulty translation. An unexpected bonus to the activity were comments from the students who were very happy to watch their own recordings just to observe what they did. It led to the very positive feedback that students believe this sort of activity to be not only entertaining but also very informative. It has to be noted that screen recording data is just a video depicting the act of translation with no further details whatsoever. Statistical results are challenging to obtain. In the case of this particular research, about 70 hours of video had to be watched an annotated. The application of eye-trackers would result in the much more detailed material to study, both for the researcher (teacher) and the students, who would draw their conclusions basing on fixations times and gaze plots.

The abovementioned research has suggested that students over-rely on CAT tools (particularly on translation memories) when translating. Eye-tracking would expose this tendency by presenting two things:

  1. the extent of use of translation memory-based matches (for matches see Bogucki, 2009; Kornacki, 2018) in CAT-based translation, including time spent on assessing their quality. Kornacki (ibid.) notes that student-translators are prone to over-rely on 100% matches, even when they are taken from external translation memory (e.g. provided by a client), frequently copying previous errors and producing poor quality translations as a result of inadequate match quality assessment. The phenomenon can be observed through two checks: i) whether the 100% match has been read at all, and ii) how much time was devoted to its evaluation. An eye-tracker will provide ample data in both cases;
  2. the extent of assessment and use of machine-translation (MT) results in CAT-based translation (MT matches are displayed in modern CAT tools alongside TM matches with different designation).

Apart from that, the eye-tracker will highlight other areas of interest to future translators:

  1. general ease of use when using the software (e.g. times needed to access a given feature);
  2. most frequently used functions (e.g. internal/external glossaries);
  3. times required to read and comprehend source text and render the target text. It includes the number of times necessary to go back to the source text, providing insights on short-term memory and related errors in translation. Séguinot (1989: 75) notes that ‘there are limitations on short-term memory. (...) We can organise our intake so the seven or so items in fact contain items themselves, for example storing words as opposed to single letters, but there is nonetheless a limit after which memory fades.’ It is especially true for CAT tools in which translators do not see context of the entire paragraph immediately but instead operate on individual translation units; and
  4. other issues as they come to attention during the data analysis process.

Group discussion allows the teacher to draw students’ attention to the abovementioned issues and come up with general advice on how the translation process should look like. For example, fixation data suggesting short time needed to read TM/MT match suggestions, considered in the light of retained TM/MT-based errors[10], would suggest inadequate match quality assessment and, as a result, over-reliance on CAT tools. Poor handling of the software would imply the need to master technical aspects of a given CAT tool before focusing on the translation process nuances. Finally, errors in translation where no matches were available would imply either problem with understanding the source text, inadequate language competence or short-term memory issues, as reported by Séguinot, for instance. She provides a most insightful comment regarding one of her studies (1989: 75):

[T]ranslator (...) made different kinds of error in the first part of sentences than towards the end of the sentences. The particular kinds of error indicate that the passage from the source text to the translation was probably through the memory of the content of the source text for the first part of the sentences, but clearly more from the surface of the source text as the translator’s memory began to fade. When this happened, there were more examples of interference from the source text, more literal translation or transcoding.

Additionally, the visual aspect of the recording is yet another boon when discussing CAT-based translation issues since students can relate issues voiced by the trainer to actual parts of the video.

3.2. Comparison of data taken from students and professionals

Comparison of translations performed by professional translators and students is a common practice both in research and in translation teaching (see, for example, Kiraly, 1995; Ehrensberger-Dow and Massey, 2013; Hunziker Heeb, 2016). Eye-tracking technology allows to take one step further and take a closer look at the variations in the process of translation as performed by professionals and student translators. General principles are similar to those in Section 3.1., i.e. the same tendencies are focused on and analysed. Differences and pinpointed and discussed.

An added value for students lies in the fact that they have unique insight into how a professional goes through the translation process. Students can see what professionals do, in what order, what tools they use, what they pay attention to. It can be achieved through replaying logfiles recorded by professionals and focusing on specific areas of interest like problem identification (linguistic or technical) and problem solving (discernable through heat maps, fixations, refixations, or regressions, for instance). In fact, the best results in terms of teaching effects can be achieved when students’ data is analysed individually first in order to hint at certain problem areas that have to be addressed and then juxtaposing it with the professional’s data in order to show the student how the process looked like, what was focused upon, and highlight differences between the two versions. Students can analyse differences in areas of interest, fixations, gaze plots and gaze duration in order to determine variations between professional and trainee translation process (see also Pietrzak and Kornacki, forthcoming).

Of course, professional-trainee performance cross analysis would be difficult time-wise (individual data capture and analysis). Therefore, group analysis may be a better choice, ie. the teacher selects 2-3 data samples from students and juxtaposes them with the data taken from professionals as described above.

The involvement of the professional translator in the process, although obviously the best option since teacher’s words and remarks are backed up by actual data, is difficult to provide due to financial issues. Author’s experience shows that it is challenging to get an experienced professional translator take part in academic activities, mainly due to time constraints and financial issues. Professional activity demands short deadlines and availability on the part of the translator and, therefore, it is virtually impossible to capture professional data ad hoc. A session has to be arranged in advance, and if the translator cannot come due to external issues (illness, professional emergency), the overall eye-tracking plan can be put under question.

Another issue is that data capture takes time which a professional would likely spend translating, i.e. earning money. Therefore, remuneration for the time spent translating (eye-tracking) would have to be provided. While this can be done easily in case of funded researches, it is more difficult in case of regular academic classes. A solution to the problem may lie in turning to fellow academics who also work as translators to take part in the project. However, not always such a person is available and if s/he is – it may cause concern for the validity of the data (the academic being also a teacher).

As can be seen from the above, both approaches are valid, with the comparison of students’ and professional’s data probably significantly harder to perform. However, new strategies are necessary, especially when we consider that our civilisation is entering the third industrial revolution, ‘one based in technological advances in software, hardware and telecommunications’ which, among others, ‘are transforming commercial practices’ (Smith, 2001: 1). For that reason, translation trainers have to meet new challenges of the developing market, discussed in the following section.

4. New challenges for translation trainers

Teaching translation has been debated by numerous researches over the years, notably Toury (1984), Kiraly (1995, 2000), Campbell (2002), González-Davies (2004), or Pym (2009). Even today, the general tendency for translator training is for the teacher to sit in front of a group of students, choose a text to translate and provide feedback (to a group or individual students) once the translation is ready (Kiraly, 2014). It is closely related to the fact that virtually no university-level programmes for translation trainers exist in the world (ibid.). Growing competition, globalisation and omnipresent technology demand revisiting this approach, especially because it has become increasingly difficult to set new learning objectives for students. Not all students choose translation track to become translators. Some choose it as a way to greatly improve their English; other saw no other reasonable option (see Kornacki, 2018). Therefore, a pro-student oriented teacher cannot set the same learning objectives for everyone. Instead, they should be individualised which is not always possible. The problem can be addressed by developing self-regulatory skills in students. Zimmerman (2000: 14) defines self-regulation as ‘self-generated thoughts, feelings, and movements that are arranged and cyclically adapted to the fulfilment of personal goals.’ Instead of pre-set goals, students need to be given a set of tools allowing them to define their own goals. Only then can they achieve success in professional life. Thus, the role of the teacher is focused on providing students with insights on the current situation on the market in terms of employment options, competition, business relations and, in case of translators, pros and cons of running own business enterprise.

One of the most important areas that have to be addressed by teachers is the use of technology in translator’s workshop. The fact that current generation of students (as of 2018) has grown up surrounded by technology and digital media does not mean that they can use them efficiently in a professional environment. Frequently they have bad, ineffective habits that have to be rid of (e.g. using mobile versions of online bilingual-dictionaries while using a PC workstation for translation). On the other hand, they lack necessary skills in text editing, bread and butter of contemporary translation. The role of the teacher is not so much as to teach students how to handle needed software (word-processors, optical-character recognition tools, basic graphics editing) but to show students that they need those skills in order to succeed as a translator. Teaching has become secondary (albeit still very important) to presenting and justifying practical requirements for the job. It includes CAT tools, mentioned before. A necessity in many cases, CAT tools are tremendous assets for many translators and translation agencies which, at the same time, set traps for the unwary. Hence the need for teachers who are (or have been) also practitioners of the trade since it is them who know where, when and what problems to expect when utilising CAT tools. Hence the idea, proposed in this paper, to use eye-trackers to show students specific areas of interest during the process of translation, not only to instruct but also to show certain tendencies in general and to stimulate individual self-development of students.

The above is inextricably linked to the fact that the pace of developments in technology and changes in the market dictate the course of action of translation (if not all) teachers. In author’s opinion, it is the greatest challenge for translation trainers – to remain one step ahead of the market and prepare students (both in terms of hard skills and mental attitude – belief in own skills and ability to adapt to the market) for professional life after graduation. Granted, it is not an easy task, especially considering limitations set by individual curricula and institutions, but it is one (if not the only) of the options to train conscious and well-prepared future translators.

5. Concluding remarks

The purpose of this paper was to share some insights on the use of the eye-tracking technology in translation classroom. Translation trainers have to follow current trends in the market to stay effective as teachers. The use of the latest technological developments (e.g. social media, see, for example, Desjardins, 2017) is seen as an important asset in contemporary classroom, not only because of their undeniable effectiveness but also due to the mind-set of contemporary students who, as was mentioned before, are so used to technology around them that they expect it also at the university. Therefore, the author believes that the use of eye-trackers[11], not as a tool of academic research but as a mean to instruct students and help develop their self-regulatory skills and better prepare them to work in certain conditions (e.g. CAT environment), is not only methodically valid but also extremely interesting to students.


Alves, Fábio, Adriana Pagano, and Igor da Silva (2009) ‘A New Window on Translators’ Cognitive Activity: Methodological Issues in the Combined Use of Eye Tracking, Key Logging and Retrospective Protocols’ in Mees, Inger M and Fábio Alves and Susanne Göpferich (eds.), Methodology, Technology and Innovation in Translation Process Research. Copenhagen: Samfundslitteratur: 267–292.

Alves, Fábio, Adriana Pagan, and Igor da Silva (2011) ‘Towards an Investigation of Reading Modalities in/for Translation: an Exploratory Study Using Eye-Tracking Data’ in O’Brien, Sharon (ed.), Cognitive Explorations of Translation. London: Continuum: 175–196.

Balling, Laura Winther; Kristian Tangsgaard Hvelplund, and Annette Camilla Sjørup (2014) ‘Evidence of Parallel Processing During Translation’, Meta 59/2, Volume 59, Number 2: 233-466

Barbu, Eduard (2015) ‘Spotting False Translation Segments in Translation Memories’ in Constantin Orăsan and Rohit Gupta (eds.), Proceedings of the Workshop on Natural Language Processing for Translation Memories (NLP4TM): 9–16. URL: (Accessed on: 11/02/2018)

Bogucki, Łukasz (2009) Tłumaczenie wspomagane komputerowo. Warszawa: PWN.

Campbell, Stuart (2002) ‘Translation in the Context of EFL: the Fifth Macroskill?’ TEFLIN Journal : Teaching English as a Foreign Language in Indonesia 13(1): 58-72.

Desjardins, Renée (2017) Translation and Social Media. In Theory, in Training and in Professional Practice. Palgrave Macmillan UK. DOI: 10.1057/978-1-137-52255-9

Dragsted, Barbara, and Inge Gorm Hansen (2008) ‘Comprehension and Production in Translation’ in Göpferich, Susanne, Arnt L. Jakobsen and Inger M. Mees (eds.), Looking at Eyes: Eye-Tracking Studies of Reading and Translation Processing. Copenhagen: Samfundslitteratur: 9–30.

Dragsted, Barbara and Michael Carl (2013) ‘Towards a Classification of Translation Styles Based on Eye-Tracking and Keylogging Data.’ Journal of Writing Research 5:1: 133–158.

Duchowski, Andrew T. (2007) Eye Tracking Methodology: Theory and Practice. London: Springer. ISBN: 1846286085

Eckstein, Maria K., Guerra-Carrillo, Belén, Miller Singley, Alison T., and Silvia A. Bunge (2017) ‘Beyond Eye Gaze: What Else Can Eyetracking Reveal About Cognition and Cognitive Development?’ Developmental Cognitive Neuroscience, Volume 25: 69-91

Ehrensberger-Dow, Maureen, and Gary Massey (2011) ‘Commenting on Translation: Implications for Translator Training.’ The Journal of Specialised Translation, vol. 16: 26-41

Ehrensberger-Dow, Maureen, Susanne Göpferich, and Sharon O’Brien (eds.) (2015) Interdisciplinarity in Translation and Interpreting Process Research. Amsterdam: John Benjamins.

Feinberg, Richard, and Edward Podolak (1965) Latency of Pupillary Reflex to Light Stimulation and Its Relationship to Aging. Washington, D.C.: Federal Aviation Agency. URL: [url=][/url] (Accessed on: 09/02/2018)

González Davies, M. (2004) Multiple Voices in the Translation Classroom. Activities, Tasks and Projects. Amsterdam and Philadelphia: Benjamins.

Göpferich, Susanne (2009) ‘Towards a Model of Translation Competence and its Acquisition: The Longitudinal Study Transcomp’ in Susanne Göpferich, Arnt L. Jakobsen, and Inger M. Mees (eds.), Behind the Mind. Methods, Models and Results in Translation Process Research. Copenhagen: Samfundslitteratur Press: 11-37.

Göpferich, Susanne (2010) ‘Data Documentation and Data Accessibility in Translation Process Research.’ The Translator 16(1): 93-124.

Harezlak, Katarzyna, Kasprowski, Paweł, and Mateusz Stasch (2014) ‘Towards Accurate Eye Tracker Calibration – Methods and Procedures.’ Procedia Computer Science 35(2014): 1073–1081

Hering, Ewald (1879) ‘Der Raumsinn und die Bewegungen des Auges [Spatial sense and movements of the eyes]’ in Herrmann, L. (ed.) Handbuch der Physiologie der Sinnesorgane [Handbook of the physiology of the sense organs] Vol. 3, Leipzig: Vogel: 341–601. Translated as ‘Spatial sense and movements of the eyes’ (C. A. Radde, Trans.). Baltimore, MD: American Academy of Optometry.

Holmqvist, Kenneth, Nyström, Marcus, Andersson, Richard, Dewhurst, Richard, Jarodzka, Halszka, and Joost van de Weijer (2011) Eye Tracking: A Comprehensive Guide to Methods and Measures. Oxford University Press.

Hunziker Heeb, Andrea (2016) ‘Professional Translators’ Self-Concepts and Directionality: Indications from Translation Process Research.’ The Journal of Specialised Translation, vol. 25: 74-88.

Hvelplund, Kristian T. (2011) Allocation of Cognitive Resources in Translation: an Eye-Tracking and Key-Logging Study. PhD thesis. Copenhagen: Copenhagen Business School. ISBN 87-593-8464-0

Hvelplund, Kristian T. (2014) Eye-tracking and the Translation Process: Reflections on the Analysis and Interpretation of Eye-tracking Data. MonTI Special Issue – Minding Translation, 2014. 201-223. ISSN 1889-4178

Jakobsen, Arnt L., and Kristian T. Hvelplund Jensen (2008) ‘Eye Movement Behaviour Across Four Different Types of Reading Task” in Göpferich, Susanne; Arnt L. Jakobsen and Inger M. Mees (eds.) Looking at Eyes: Eye-Tracking Studies of Reading and Translation Processing. Copenhagen: Samfundslitteratur: 103–124.

Just, Marcel A. and Patricia A. Carpenter (1980) ‘A Theory of Reading: From Eye Fixations to Comprehension.’ Psychological Review 87: 329-354.

Kiraly, Don (1995) Pathways to Translation: Pedagogy and Process. Kent, Ohio, Kent State University Press.

Kiraly, Don (2000) A Social Constructivist Approach to Translator Education. Manchester: St. Jerome.

Kiraly, Don (2014) ‘From Assumptions about Knowing and Learning to Praxis in Translator Education.’ inTRAlinea Special Issue: Challenges in Translation Pedagogy. URL: [url=][/url] (Accessed on: 12/02/2018)

Kornacki, Michał (2018) Computer-assisted Translation (CAT) Tools in the Translator Training Process. Frankfurt am Main: Peter Lang.

Martinez-Conde, Susana, Macknik, Stephen L., Troncoso, Xoana G., and David H. Hubel (2009) ‘Microsaccades: a Neurophysiological Analysis.’ Trends in Neurosciences, Volume 32, Issue 9, September 2009: 463-475

O’Brien, Sharon (2006) ‘Pauses as Indicators of Cognitive Effort in Post-Editing Machine Translation Output.’ Across Languages and Cultures 7 (1): 1-21

O’Brien, Sharon (2009) ‘Eye Tracking in Translation Process Research: Methodological Challenges and Solutions’ in Susanne Göpferich, Fabio Alves, and Inger M. Mess (eds.), Methodology, Technology and Innovation in Translation Process Research: a Tribute to Arnt Lykke Jakobsen. Copenhagen: Samfundslitteratur: 251-266.

O’Brien, Sharon (2015) ‘The Borrowers. Researching the Cognitive Aspects of Translation’ in Ehrensberger-Dow, Maureen, Susanne Göpferich, and Sharon O’Brien (eds.). Interdisciplinarity in Translation and Interpreting Process Research. Amsterdam: John Benjamins: 5-17.

Pavlović, Nataša, and Kristian T. Hvelplund Jensen (2009) ‘Eye Tracking Translation Directionality” in Pym, Anthony and Alexander Perekrestenko (eds.), Translation Research Projects 2. Universitat Rovira i Virgili, Tarragona: 101–119.

Pietrzak, Paulina, and Michał Kornacki (forthcoming) ‘The Application of New Technologies in Translator Training; Eye-Tracking Data as a Window To Translators’ Actual Actions’ in Bogucki, Łukasz, Pietrzak, Paulina, and Michał Kornacki (eds.), Translator Professional Development. Frankfurt am Main: Peter Lang.

Pym, A. (2009) Translator Training. Pre-print Text Written for the Oxford Companion to Translation Studies. URL: [url=][/url] (Accessed on: 07/02/2018).

Rayner, Keith (1998) ‘Eye Movements in Reading and Information Processing: 20 Years of Research.’ Psychol Bull. 1998 Nov;124(3): 372-422.

Saldanha, Gabriela and Sharon O’Brien (2013) Research Methodologies in Translation Studies. London and New York: Routledge

Séguinot, Candace (1989) ‘Understanding Why Translators Make Mistakes.’ TTR : traduction, terminologie, rédaction, 2(2): 73–81. doi:10.7202/037047ar

URL: [url=][/url] (Accessed on: 11/02/2018)

Sjørup, Annette Camilla. (2013) Cognitive Effort in Metaphor Translation: An Eye-tracking and Key-logging Study. PhD thesis. Copenhagen: Copenhagen Business School.

Smith, Bradford L. (2001) ‘The Third Industrial Revolution: Policymaking for the Internet.’ The Columbia Science and Technology Law Review, Volume III (2001-2002): 1-45. URL: [url=][/url] (Accessed on: 11/02/2018)

Teixeira, Carlos (2014) Data Collection Methods for Researching the Interaction Between Translators and Translation Tools—an ‘Ecological’ Approach’ in Schwieter, John, and Aline Ferreira (eds.), The Development of Translation Competence: Theories and Methodologies from Psycholinguistics and Cognitive Science. Cambridge: Cambridge Scholars Publishing: 269-286.

Tobii (2018) ‘Light Conditions.’ URL: [url=][/url] (Accessed on: 07/02/2018)

Toury, Gideon (1984) ‘The Notion of “Native Translator” and Translation Teaching’ in W. Wilss, and G. Thome (eds.), Die Theorie des Übersetzens und ihr Aufschlusswert für die Übersetzungs- und Dolmetschdidaktik. Tübingen: Narr: 186-195.

van der Lans, Ralf, Wedel, Michel, and Rik Pieters. 2011. ‘Defining Eye-Fixation Sequences Across Individuals and Tasks: The Binocular-Individual Threshold (Bit) Algorithm.’ Behav Res Methods, 43(1): 239–257.

van Eijsden, Pieter, Fahmeed Hyder, Douglas L. Rothman, and Robert G. Shulman (2009) ‘Neurophysiology of Functional Imaging.’ Neuroimage 45(4): 1047–1054.

Wade, Nicolas J. 2015. ‘How Were Eye Movements Recorded Before Yarbus?’ Perception 44(2015): 851-883. DOI:10.1177/0301006615594947

Wells, William C. 1792. An Essay upon Single Vision with Two Eyes: Together with Experiments and Observations on Several Other Subjects in Optics. London: Cadell.

Zimmerman, Barry J. 2000. ‘Attaining Self-regulation: A Social Cognitive Perspective’ in Boekaerts, Monique, Paul R. Pintrich and Moshe Zeidner (eds.), Handbook of Self-regulation. San Diego, CA: Academic Press: 13–39. [url=][/url]


[1] As opposed to machine translation, which is not the subject of the herein paper.

[2] One of its aspects, functional imaging, can be used, for example, in cognitive psychology research (van Eijsden, Hyder, Rothman and Shulman, 2009).

[3] In order to use eye-tracking successfully in the translation classroom, the role of the trainer has to be regarded as similar to that of the researcher in the respect that a unique type of data has to be collected in order to allow students benefit from it. Therefore, specific goals have to be set and the process designed in detail (the research part). Once the data is obtained, it can be used to stimulate learning process of the students (teaching part).

[4] In case they do, more data capture methods may be (and actually should) used, e.g. retrospective TAPs or webcam recording of the translator.

[5] Mean duration of fixation for silent reading is 225-250 ms and 275-300 ms for reading out loud (see Rayner, 1998; van der Lans, Wedel, and Pieters, 2011; Hvelplund, 2014). In case of typing (involving simultaneous typing and reading), the mean duration is 400 ms. Visual search is reported to last between 180–275 ms and scene viewing between 260-330 ms (van der Lans, Wedel, and Pieters, 2011). As a result, it is ‘important to interpret the fixation data in light of the kind of reading that the translator is performing’ Hvelplund (2014: 213).

[6] ‘Gaze plots show the location, order, and time spent looking at locations on the stimulus, whether web page, print advertisement, or video. So the primary function of the gaze plot is to reveal the time sequence of looking or where we look and when we look there. Time spent looking, most commonly expressed as fixation duration, is shown by the diameter of the fixation circles. The longer the look, the larger the circle’ (Tobii, 2018: internet resource).

[7] ‘Heat maps show how looking is distributed over the stimulus. [They] are a visualization that can effectively reveal the focus of visual attention for dozens or even hundreds of participants at a time’ (Tobii, 2018: internet resource). Unlike gaze plots, they do not provide any information about the order of looking and they do not focus on individual fixations.

[8] Considering 1.5 hours of computer-lab translation classes a week

[9] Regressions and refixations may suggest a number of problems in CAT environment, e.g., issues with segmentation, processing of tags, understanding of content, issues with match quality (both TM- and MT-based) or terminology problems. What is more, they may suggest software handling issues and point to most commonly used features during a standard translation workflow.

[10] For more information on TM-based errors please refer to Barbu (2015) and Kornacki (2018).

[11] Or at least screen recorders, considering limited availability of eye-tracking in public institutions.

About the author(s)

Michał Kornacki, PhD, is an Assistant Lecturer in Translation Studies at the Department of Translation, Institute of English Studies, University of Łódź, Poland. His primary area of research is translator education. He focuses on the use of technology in translator’s workshop, including computer-assisted translation (CAT) and audiovisual translation (AVT) tools. Before starting academic career, Michał worked as translation project manager for 7 years which gave him deep understanding of the translation market in Poland.He is responsible f or implementing of CAT programme at the Institute of English Studies. Since 2017 he is a certified Memsource trainer.

Email: [please login or register to view author's email address]

©inTRAlinea & Michał Kornacki (2019).
"The application of eye-tracking in translator training"
inTRAlinea Special Issue: New Insights into Translator Training
Edited by: Paulina Pietrzak
This article can be freely reproduced under Creative Commons License.
Stable URL:

Go to top of page