2. Early eye-tracking research with bilinguals (1997 – present)

In the late 1990s and early 2000s, eye-tracking research on bilingualism was still a niche area compared to the extensive research with so-called monolinguals already being conducted in cognitive psychology and psycholinguistics (for landmark reviews, see Rayner, Reference Rayner1998, Reference Rayner2009; Tanenhaus & Trueswell, Reference Tanenhaus, Trueswell, Traxler and Gernsbacher2006). Two major paradigms were influential at the time and remain so to this date, namely, eye tracking with reading and the visual world eye-tracking paradigm.Footnote ¹ Some of the pioneering eye-tracking research in these two paradigms that we would now consider to mark the beginning of eye tracking in SLA and bilingualism originated in the realms of psychology (Marian & Spivey, Reference Marian and Spivey2003a, Reference Marian and Spivey2003b; Spivey & Marian, Reference Spivey and Marian1999) and linguistics (Frenck-Mestre & Pynte, Reference Frenck-Mestre and Pynte1997), highlighting its interdisciplinary beginnings.

Researchers at the time used eye-movement registration to examine questions about linguistic representation or knowledge, often as it relates to morphology and syntax (e.g., Keating, Reference Keating2009), and sometimes, the lexicon (Marian & Spivey, Reference Marian and Spivey2003a, Reference Marian and Spivey2003b; Spivey & Marian, Reference Spivey and Marian1999). Another focus was crosslinguistic influence and L2-specific factors in parsing, which is the mental process of building and revising syntactic structures to enable language comprehension (Dussias & Sagarra, Reference Dussias and Sagarra2007; Frenck-Mestre & Pynte, Reference Frenck-Mestre and Pynte1997; Roberts et al., Reference Roberts, Gullberg and Indefrey2008). Although eye movements by definition are recorded while participants are performing a task, that is, in real time during processing, the theoretical focus was on the cognitive architecture supporting language acquisition that the eye movements were believed to reflect; that is, eye tracking was primarily a knowledge measure. The materials (sentences) that participants read or listened to were not intended to teach them new information about the language. They were not meant as an instructional intervention (and hence, we will not consider them further in our ISLA timeline), but rather, they were specifically designed to examine linguistic representations (existing knowledge) and parsing procedures that were already established in the bilingual mind as a result of exposure or prior instruction. This idea is sometimes referred to as the “processing outcomes of acquisition” (e.g., Godfroid, Reference Godfroid and Leow2019; VanPatten & Rothman, Reference VanPatten, Rothman and Rebuschat2015). It would be another decade before eye-tracking researchers in SLA, (psycho)linguistics, and applied linguistics discovered that eye tracking could also be used to study how processing can lead to acquisition and started exploring this powerful idea – termed “acquisition-as-processing” (e.g., Godfroid, Reference Godfroid and Leow2019; VanPatten & Rothman, Reference VanPatten, Rothman and Rebuschat2015) – in ISLA research.

3. Eye tracking as a tool to study acquisition-as-processing (2010 – present)

The second wave of eye-tracking research in SLA began around 2010, when eye-movement registration was newly introduced as a processing measureFootnote ² (Godfroid et al., 2013*; Smith, 2013*). This innovative use of eye tracking empowered researchers to capture learning processes in real time and measure the effects of an instructional intervention designed to promote language learning (see Lai et al., Reference Lai, Tsai, Yang, Hsu, Liu, Lee, Lee, Chiou, Liang and Tsai2013, for a similar review in education on learning-related eye-tracking research). As such, the second wave of eye tracking marked the beginning of eye tracking in ISLA (also see Godfroid, Reference Godfroid and Leow2019, for a review).

Two foundational assumptions informed eye-tracking research in ISLA, namely the eye-mind link (Just & Carpenter, Reference Just and Carpenter1980; Reichle et al., Reference Reichle, Pollatsek and Rayner2006) and Schmidt’s noticing hypothesis (e.g., Schmidt, Reference Schmidt1990). The eye-mind link in applied eye-tracking research is the notion that the eye gaze, as a marker of overt attention, is linked with ongoing cognitive processing; in other words, overt and covert attention are connected, even if the strength of the connection can vary. Attention, of course, is a central component of Schmidt’s noticing hypothesis, which highlights a crucial role for focal attention and a low level of awareness, called noticing, in second language learning. Combined, these two assumptions make eye-movement registration an exquisite tool to examine attentional allocation in language learning.

One line of related language-learning research has examined the role of attention in vocabulary learning, using both unimodal (reading-only) and multimodal (reading-while-listening, captioned/subtitled videos) input. Results generally support the beneficial role of attention, especially for the acquisition of meaning, and especially in reading-only contexts. The more attention a reader pays to an unfamiliar word during reading (i.e., the longer they look at it, as measured with an eye-tracking camera), the more likely they are to recognize or recall that word on a surprise test after the reading (e.g., Elgort et al., 2018*; Godfroid et al., 2013*, 2018*; Pellicer-Sánchez, 2016*). In multimodal studies with captioned/subtitled videos, the attention-learning relationship appears to be more complicated (e.g., Montero Perez et al., 2015*), potentially due to the need for learners to integrate both auditory and visual information in real time. Understanding how L1 and L2 users are able to accomplish this complex task successfully is another focus of contemporary eye-movement research (e.g., for reading-while-listening: Conklin et al., 2020*; for captioned/subtitled videos: Liao et al., 2022*).

The basic question of an attention-learning relationship is now being examined across a range of learning conditions, from implicit (unsupported, meaning-focused, input-only) input conditions to highly explicit (directed, form-focused) interventions (e.g., Indrarathne & Kormos, 2017*). Researchers have designed different forms of modified input with the goal of increasing the salience of the target structures in the input. They do this because salience may impact attention, and attention may increase the likelihood that language learners will notice and potentially learn the new target structures. Examples of input modifications are aural and visual enhancement, glossing, and different manipulations of task instructions designed to direct learners’ attention towards the form and/or meaning of the target structures (as described in the classic work by Doughty & Williams, Reference Doughty, Williams, Doughty and Williams1998, and Sharwood Smith, Reference Sharwood Smith1993). Across all of these, the novel addition of eye-movement registration, again, can tell how effective these interventions are in directing a learner’s attention to specific structures in the input. Eye tracking, therefore, provides the missing link between instructional design and student reception.

4. The shift to ecologically valid, remote, and wearable eye tracking (2018 – present)

In the past five to ten years, eye tracking has emerged as a readily accessible technology for researchers investigating second language learning in different settings, including those that do not require a computer in a lab. The current utilization of eye tracking therefore covers a full spectrum of possibilities, spanning a continuum between experimental control and ecological validity. On the one end of this continuum, we continue to see studies conducted in controlled laboratory settings that allow researchers to systematically manipulate variables, isolate specific factor(s), and strengthen causal inferences through experimental designs. In many of these cases, researchers employ highly accurate and precise, high-speed eye trackers, thus optimizing data quality, at least in principle (see Godfroid et al., Reference Godfroid, Finch and Koh2025). In contrast, the adoption of eye tracking in more naturalistic settings permits the investigation of learners engaging in authentic language learning tasks and activities. A parallel trend, driven in part by the COVID-19 global health crisis, is the shift from lab-based to remote data collection, which provides access to larger and more diverse participant pools. By leveraging webcam-based eye-tracking functionality on platforms like Gorilla, researchers are no longer confined to physical labs. These technological advances – including virtual data collection platforms and a full range of eye-tracking devices with varying technical capabilities – have eliminated the need for participants to remain tethered to a computer setup with a chinrest in an eye-tracking lab. As a result, researchers are free to explore eye-tracking applications in diverse real-world situations, thereby enhancing the external validity of the study.

These applications of eye tracking outside the lab include but are not limited to classrooms (Morell, 2018*), interactional contexts (McDonough et al., 2015*), computer-assisted language learning environments (Fievez et al., Reference Fievez, Montero Perez, Cornillie and Desmet2023; Michel & O’Rourke, 2019*; Michel & Smith, Reference Michel, Smith, Gass, Spinner and Behney2018), and online instruction (Satar & Wigham, 2017*). In some studies, eye movements remain a primary source of data; in other studies, researchers engage in intense triangulation of different research methodologies, whereby eye-gaze tracking provides an additional source of information amidst many others. In the latter case, the various data sources provide researchers with a more complete picture of the sociocognitive and affective processes that take place in L2 instructional settings (e.g., Lindberg et al., 2021*; Morell, 2018*). Given the growing interest in using eye tracking for classroom-based research on second language learning and instruction, wearable eye trackers seem to hold special promise because of their portability. To our knowledge, such research has not been attempted yet in ISLA. However, in education, researchers are actively using wearable eye trackers in classrooms, investigating areas such as the professional vision of the teacher (e.g., selective attention to a classroom situation, leading to classroom management decisions), sometimes with a focus on teacher training (Witt et al., Reference Witt, Schorer, Loffing and Roden2024). A specific research question in this line of work is the role of teaching expertise in understanding a classroom situation (Miller et al., Reference Miller, Correa, Cortina, Phelps and Chamberlain2024).

Another potential application of wearable devices lies in the ability to integrate augmented and virtual realities. This advance has considerable commercial potential, prompting tech giants like Google and Apple to acquire companies specializing in eye-tracking technology. New headsets have interactive eye-tracking applications built into them, which allow the viewer, for example, to make selections in a user interface simply by looking at the relevant objects.

As commercial applications of eye tracking become ubiquitous, we are also beginning to witness the integration of these ideas into the realm of L2 research. For instance, Révész and colleagues (2023*) used a first potential application of interactive eye tracking in vocabulary studies that demonstrated the usefulness of interactive, gaze-contingent support in instructed language learning. In their study, lexical glosses that readers would otherwise overlook (as revealed in skips of the glosses in the eye-movement records) were made visually more salient through yellow highlighting, whereas glosses that readers looked at (as revealed by fixations on the glosses in the eye-movement records) remained unenhanced. The interactive addition of enhancement – that is, only in cases where learners needed it – highlights how eye-movement registration could be integrated as a part of an instructional treatment or language assessment to tailor the input to an individual learner’s behavior. Considering the rapid, continuous pace of technological advancements, we foresee that an increasing number of pioneering studies will begin to incorporate eye tracking in these innovative, interactive applications (e.g., Smith & González-Lloret, Reference Smith and González-Lloret2021).

While these developments are exciting, researchers are reminded of the importance of the balance between ecological validity and methodological rigor. For example, by opting for commonly available technology like video cameras or webcams for recording eye movements, researchers often compromise the accuracy and precision of their recordings. This means that certain topics and questions (e.g., those requiring small interest areas, like words or units below the word level in reading) cannot be investigated in a valid manner with off-the-shelf eye-tracking devices like webcams, even though such work does get attempted sometimes. Despite the many technological advances, we believe that substantive questions should continue to drive the research process more than the technology. Eye-movement registration is simply a tool, and nothing more than that. Researchers enjoy the freedom to use that tool in ways that best suit their research purposes. They are, at the same time, presented with the challenge to judiciously select their eye-tracking equipment and match their research designs with the eye-tracker’s technical capabilities (Godfroid & Hui, Reference Godfroid and Hui2020).

In summary, eye tracking has continued to gain momentum as a valuable research method in ISLA research. Its versatility allows researchers to strike a balance between experimental control and ecological validity, catering to a wide range of research purposes.

5. Preview of research timeline

In what follows, we present a timeline of eye tracking in ISLA research. We review the historical significance of each study and situate it within the broader context of earlier and later works. Our timeline highlights the emergence and evolution of eye tracking as a tool for studying acquisition-as-processing – using eye movements to examine how attention and processing varies across individuals and learning conditions, and how these variations influence or mediate eventual learning outcomes (which are often measured independently through offline tests).

Major themes emerging from this research include investigations into incidental and intentional vocabulary learning during reading. Here, researchers have manipulated variables such as exposure frequency, input enhancement, glosses, and explicit instruction to examine their influence on attention, processing, and learning. Eye-movement registration has also been employed to evaluate and compare the effectiveness of instructional approaches, including processing instruction. Additionally, the affordances of multimodal input for language learning – including captions, reading-while-listening, and picture books – have been a focal point since the early days of eye-tracking in ISLA.

Increasingly, eye movements are being triangulated with other data sources, including stimulated recall, interviews, screen capture, keystroke logging, and skin conductance. These multi-method approaches provide rich accounts of complex activities, adding important insights to research on writing, face-to-face or computer-mediated interaction, and online or in-person teaching. We are also witnessing the emergence of eye tracking in social approaches to ISLA, such as analyzing eye gaze as a multimodal or semiotic resource in classroom or peer-to-peer interactions, or using eye tracking as a tool to promote learner development in L2 writing.

The timeline below includes 73 studies, each coded for the key characteristics outlined in Table 1.

Table 1. Codes for study characteristics