Research directions in eye-tracking methodology for second language research

Tineke Brunfaut – 25/04/2018

In recent years, we have seen an increasing use of eye-tracking methodology in second language research. One fruitful line of research has used the technology to gain insights into second language learners’ cognitive processing while completing learning and assessment tasks. Here are some examples of studies which have used eye-tracking as part of their research design to look into the nature of, for instance, task processing in second language reading, writing or video-listening. The assumption underlying the use of eye-tracking is that there is a close link between what our eyes are looking at and what the focus of our attention is at that point in time.

The increasing body of empirical work in second language learning and assessment seems to support the usefulness of eye-tracking methodology in at least certain respects. For example, there appear to be some interesting links between eye movements and second language reading ability. In research I conducted with Gareth McCray (Brunfaut & McCray, 2015a), for instance, we found that eye movement data are particularly useful to gain insights into so-called lower-level reading processes such as word decoding. On the other hand, we found that they reveal less or no information on higher-level reading processes such as inferring information from a text. In the majority of second language studies, therefore, eye-tracking methodology has been used in combination with other research methods such as retrospective interviews, to compensate for limitations in the methodology and to triangulate data.

A somewhat different line of research concerns explorations of the potential of eye-tracking methodology for modelling and assessing second language abilities. This corresponds with the increase in studies on how technology can help inform second language assessment, which is a promising avenue of research. For example, in another study (Brunfaut & McCray, 2015b), we were interested in exploring the potential usefulness of eye-tracking for modelling (the testing of) second language reading. We found that a selection of eye movement measures predicted English language learners’ reading ability to some extent, but certainly not fully. Similar research has recently been reported in Berzak, Katz & Levy (2018), who explored the relationship between eye movement data collected during sentence-level reading and performance on a number of general language proficiency tests.

As eye-tracking technology becomes less expensive and more user-friendly, and as research sheds more light on what eye movements can and cannot tell us, eye movement data might provide useful, complementary information in second language reading assessment. For example, in future, eye-tracking might help second language teachers and testers diagnose lower-level reading problems as part of a battery of assessment tools. At the same time, we need to keep in mind the limitations of what eye movements can capture: we should not overstate their potential to assess the wide range of skills and processes involved in language proficiency as a whole (e.g., inferencing while reading or listening, or speaking skills). Findings from the first line of research described above could be helpful in deciding on what are meaningful and valid associations or predictions to explore in the second line of research mentioned above. Another caveat is that effects derived under research lab conditions might not necessarily translate into operational assessment conditions. In addition, we should keep in mind that eye-tracking in this context would constitute an indirect form of comprehension assessment, whereas direct forms of assessment are desirable – not the least from the perspective of generating positive effects on language learning and teaching. Essentially, much more research is needed before any strong claims can be made about the potential of eye-tracking for assessment purposes, but it seems a worthwhile area for further exploration in some respects (e.g. lower-level reading processes).