ISLA research question 1: How are L2s learned in instructed contexts?

Following several decades of research, the question is no longer if learning an L2 occurs in instructed contexts, but how and under what conditions. Instruction, or the intentional and systematic manipulation of pedagogical approaches and empirical treatments, does not change the route of learning, but instruction can and often does change the rate and the amount of learning (e.g., Norris & Ortega, Reference Norris and Ortega2000; Li & Sun, Reference Li and Sun2024 for general meta-analyses; Pellicer-Sánchez et al., Reference Pellicer-Sánchez, Conklin and Vilkaitė-Lozdienė2021; Yu & Trainin, Reference Yu and Trainin2022 for vocabulary learning; Lee et al., Reference Lee, Jang and Plonsky2015; Zhang & Yuan, Reference Zhang and Yuan2020 for pronunciation instruction; Dalman & Plonsky, Reference Dalman and Plonsky2022 for listening strategy instruction; Bardovi-Harlig & Vellenga, Reference Bardovi-Harlig and Vellenga2012; Bardovi-Harlig et al., Reference Bardovi-Harlig, Mossman and Vellenga2015; Plonsky & Zhuang, Reference Plonsky, Zhuang and Taguchi2019 for pragmatics instruction). Thus, a first principal research question in ISLA addresses the variables that impact L2 learning in instructed contexts. Some of the commonly investigated variables include the provision, type, and timing of corrective feedback (Fu & Li, Reference Fu and Li2022; Lyster et al., Reference Lyster, Saito and Sato2013 for oral feedback; Brown et al., Reference Brown, Liu and Norouzian2023 for written feedback; Canals et al., Reference Canals, Granena, Yilmaz and Malicka2021 for computer-mediated feedback), type and/or timing of instruction (Michaud & Ammar, Reference Michaud and Ammar2023; Tsai, Reference Tsai2020; Umeda et al., Reference Umeda, Snape, Yusa and Wiltshire2019 for effect of explicit instruction on grammar; Ahmadian, Reference Ahmadian2020 for comparison of explicit vs implicit on instruction of pragmatics; Kang et al., Reference Kang, Sok and Han2019; Li & Sun, Reference Li and Sun2024 for meta-analyses of explicit vs implicit instruction), and the modality of instruction (Peterson, Reference Peterson2021 for impact on speaking proficiency; Hiromori, Reference Hiromori2023 for the effect on group dynamics; Awada & Diab, Reference Awada and Diab2023; Jiang et al., Reference Jiang, Kalyuga and Sweller2021 for the impact on writing; Moradi & Farvardin, Reference Moradi and Farvardin2020 on the effect on negotiation of meaning; Yousefi & Nassaji, Reference Yousefi and Nassaji2019 for meta-analyses of face-to-face vs computer-mediated instruction; Dixon et al., Reference Dixon, Christison, Dixon and Palmer2021 for meta-analysis of hybrid language instruction). There is nothing unique about the mechanism of L2 learning that occurs in different instructional contexts; it is the same whether it takes place in an instructed or non-instructed, incidental, or more naturalistic setting (such as work abroad or a romantic partnership, just to name two examples). Rather, it is the speed of L2 learning and the depth and breadth of linguistics knowledge (pragmatics, vocabulary, grammar, pronunciation) and mastery of different skills areas (listening, reading, writing, speaking) that are influenced by the variables in each setting and each learner.

ISLA research question 2: What is the nature of the L2 knowledge gained in instructed contexts?

A second principal ISLA research question (ISLA RQ) targets the nature of L2 knowledge that is gained through instructed contexts. Thus far, overall studies have found explicit knowledge to be the primary type of knowledge obtained in instructed contexts, due to the emphasis on explicit instruction in the classroom (e.g. Goo et al., Reference Goo, Granena, Yilmaz, Novella and Rebuschat2015; Leow, Reference Leow2015b; Norris & Ortega, Reference Norris and Ortega2000), which is often selected based on the limited time in classroom and the expectations of what instruction is realistic (Graus & Coppen, Reference Graus and Coppen2016, Reference Graus and Coppen2017; Mansouri et al., Reference Mansouri, Jami and Salmani2019; Nassaji, Reference Nassaji2012; Sato & Oyanedel, Reference Sato and Oyanedel2019; Schurz & Coumel, Reference Schurz and Coumel2023). Despite this dominance of explicit instruction in the L2 classroom, research has also found that explicit learning can become more automatized and used spontaneously with practice, and that explicit knowledge can occur from implicit learning (Ellis, Reference Ellis, Ellis, Loewen, Elder, Reinders, Erlam and Philp2009; Ellis et al., Reference Ellis, Loewen, Elder, Reinders, Erlam and Philps2009; Leow, Reference Leow2000; Suzuki & DeKeyser, Reference Suzuki and DeKeyser2017; for a more theoretical discussion, see Hulstijn, Reference Hulstijn2002). While not as common, implicit knowledge can also result from instructed contexts (Akakura, Reference Akakura2012; Khezrlou, Reference Khezrlou2021; Spada & Tomita, Reference Spada and Tomita2010; Williams, Reference Williams, VanPatten, Williams, Rott and Overstreet2004, Reference Williams2005). Given this, and the fact that implicit knowledge has been found to outlast explicit knowledge when studies have used delayed posttests (Goo et al., Reference Goo, Granena, Yilmaz, Novella and Rebuschat2015; Kang et al., Reference Kang, Sok and Han2019), considering both explicit and implicit learning is recommended in instructed contexts (Leow, Reference Leow2015b; Loewen, Reference Loewen2020a). The specifics regarding instructional techniques for each competency and skill must be considered in tandem with contextual variables (RQ3) and learners’ individual differences (IDs) (RQ4).

ISLA research question 3: How do variables related to instructed contexts influence L2 learning?

A third principal research question examines how key variables influence the L2 learning process and learning outcomes. For example, variables such as input (Gass et al., Reference Gass, Mackey and Pica1998; Gurzynski-Weiss et al., Reference Gurzynski-Weiss, Geeslin, Daidone, Linford, Long, Michalski, Solon, Tyler, Ortega, Uno and Park2018; Krashen, Reference Krashen1985), attention (Leow et al., Reference Leow, Thinglum and Leow2022; Schmidt, Reference Schmidt and Schmidt1995), interaction (Long, Reference Long2020; Mackey, Reference Mackey1999; Mackey et al., Reference Mackey, Gass and McDonough2000), output (Izumi et al., Reference Izumi, Bigelow, Fujiwara and Fearnow1999; Swain & Lapkin, Reference Swain and Lapkin1998; Zalbidea, Reference Zalbidea2021), as well as meaningful contexts of use (Ellis et al., Reference Ellis, Tanaka and Yamazaki1994; Long, Reference Long2015), amount of time (Tracy-Ventura et al., Reference Tracy-Ventura, Huensch, Mitchell, Le Bryun and Paquot2021), number and type of interlocutors (Gurzynski-Weiss, Reference Gurzynski-Weiss2020; Gurzynski-Weiss & Plonsky, Reference Gurzynski-Weiss, Plonsky and Gurzynski-Weiss2017), and diversity of L2 opportunities (Baker et al., Reference Baker, Martinsen, Dewey, Brown and Johnson2010; Pérez-Vidal & Juan-Garau, Reference Pérez-Vidal and Juan-Garau2011) have all been found to influence the nature of L2 learning. Adding to this already impressive constellation of factors, each of these variables are further differentiated in presence and impact depending on the context of instruction – “foreign” vs. second language contexts (Borràs & Llanes, Reference Borràs and Llanes2022; Collentine & Freed, Reference Collentine and Freed2004; Felder & Henriques, Reference Felder and Henriques1995), heritage vs. second language contexts (Bowles et al., Reference Bowles, Adams and Toth2014), immersion abroad vs. domestic immersion vs. classroom-only instruction (Freed et al., Reference Freed, Segalowitz and Dewey2004; Serrano et al., Reference Serrano, Llanes and Tragant2011), and use of digital language learning apps (Kessler et al., Reference Kessler, Loewen and Trego2020; Loewen et al., Reference Loewen, Crowther, Isbell, Kim, Maloney, Miller and Rawal2019, Reference Loewen, Isbell and Sporn2020). There is no straightforward answer nor is there space in this nor any other single article to discuss the nuances of the myriad variables at play in instructed L2 contexts; here I will simply draw awareness to the fact that each context has many variables at play, and while some are expected across contexts (e.g., a certain amount of exposure in university-level L2 classes), others vary widely (e.g., amount of L2 exposure during study abroad) and, critically, context and within-context variables are experienced differentially by individual students (see principle research question four that follows) and can be experienced differentially by the same person in different ways at different times.

ISLA research question 4: How do IDs play a role in instructed L2 learning?

IDs, or characteristics that we all have, and which we use to differentiate between and compare across individuals and groups, are the focus of a fourth principal research question in ISLA. IDs are used as data points to assist in understanding how L2 learning occurs, what can be done to maximize L2 learning opportunities, and to better understand that certain actions may be differentially beneficial for specific learners and/or groups of learners (Albert & Csizér, Reference Albert and Csizér2022; Al-Hoorie, Reference Al-Hoorie2018; Botes et al., Reference Botes, Dewaele and Greiff2020, Reference Botes, Dewaele and Greiff2022; Chen et al., Reference Chen, He, Swanson, Cai and Fan2022). As currently conceived, there are at least four types of IDs in SLA research: sociocultural and demographic IDs, such as age (Jaekel et al., Reference Jaekel, Schurig, Florian and Ritter2017; Singleton & Pfenninger, Reference Singleton, Pfenninger, S., Hiver and Papi2022); cognitive IDs, such as working memory (Jackson, Reference Jackson2020; Li et al., Reference Li, Ellis and Zhu2019; Wen & Jackson, Reference Wen, Jackson, S., Hiver and Papi2022); conative IDs, such as willingness to communicate (Mystkowska-Wiertelak, Reference Mystkowska-Wiertelak2021; Peng, Reference Peng, S., Hiver and Papi2022); and affective IDs, such as anxiety (Botes et al., Reference Botes, Dewaele and Greiff2020; MacIntyre & Wang, Reference MacIntyre, Wang, S., Hiver and Papi2022; Zhang, Reference Zhang2019); for volume-length treatments of IDs in the larger field of SLA, see Li et al. (Reference Li, Hiver and Papi2022b), among others; for ISLA-specific treatments of IDs, see Li (Reference Li2024) and Gurzynski-Weiss (Reference Gurzynski-Weiss2017a, Reference Gurzynski-Weiss2020).

The earliest studies in the larger field of SLA focused on the IDs that “good language learners” possessed, such as a willingness to take risks and guess, a willingness to communicate, being unconcerned with making mistakes, looking for patterns in form, seeking out opportunities to practice, monitoring their own and others’ speech, and paying attention to meaning (Ruben, Reference Ruben1975). More recently, in both SLA and ISLA studies, individual IDs such as personality (Anggraini et al., Reference Anggraini, Cahyono, Anugerahwati and Ivone2022; Chen et al., Reference Chen, He, Swanson, Cai and Fan2022), language aptitude (Benson & DeKeyser, Reference Benson and DeKeyser2019; Wen & Skehan, Reference Wen and Skehan2021), motivation (Oga-Baldwin & Nakata, Reference Oga-Baldwin and Nakata2017; Yousefi & Mahmoodi, Reference Yousefi and Mahmoodi2022), learning and/or cognitive styles (Griffiths & İnceçay, Reference Griffiths and İnceçay2016; Sun & Teng, Reference Sun and Teng2017), and learning strategies and self-regulation (Sun & Wang, Reference Sun and Wang2020; Teng, Reference Teng2024; Zhang et al., Reference Zhang, Thomas and Qin2019), had been considered as static variables (or, at least, were measured as such) and in relationship to learning processes and outcomes (Dörnyei & Ryan, Reference Dörnyei and Ryan2015). IDs have since been (re)conceptualized as dynamic and are considered and measured empirically in concert with each other and the contexts in which they are measured (Gurzynski-Weiss, Reference Gurzynski-Weiss2020; Serafini & Sanz, Reference Serafini and Sanz2016; see also the 2023 issue on L2 anxiety from this journal). Additionally, more affective IDs have been considered in the mix, such as well-being (Pan et al., Reference Pan, Wang and Derakhshan2023), enjoyment (Dewaele et al., Reference Dewaele, Saito and Halimi2023; Resnik et al., Reference Resnik, Dewaele and Knechtelsdorfer2023), boredom (Li et al., Reference Li, Dewaele and Hu2023; Pawlak et al., Reference Pawlak, Kruk, Zawodniak and Pasikowski2020), among others. However, the majority of published research (until the mid-2010s) has relied on instruments that reflected this fixed viewpoint; more on this in Unique Methodological Challenges in ISLA Research below.

ISLA research question 5: What instructional techniques are most likely to facilitate ISLA?

A fifth principal research question targets how to use the information from the first four questions to facilitate ISLA. Like all worthy questions, there is no “quick” answer – it depends on the theoretical framework, the methods chosen, the individual student(s), the context, as well as the skill/competency area, and so on, among other considerations. When beginning an ISLA research study, meta-analyses are a useful starting point to understand broadly what has been found to be effective for a given type of manipulation (see Plonsky’s public record of meta-analyses published in the field of applied linguistics; Plonsky, Reference Plonskyn.d.). For example, finding that implicit instruction maybe more lasting than explicit instruction (Kang et al., Reference Kang, Sok and Han2019; but see Norris & Ortega, Reference Norris and Ortega2000) can provide a valuable starting point for a more nuanced examination of instructional interventions in relationship to specific skills or competencies (for vocabulary, see Al-Hoorie et al., Reference Al-Hoorie, Vitta and Nicklin2023; Güvendir et al., Reference Güvendir, Borràs and Güvendir2024; Hao et al., Reference Hao, Wang and Ardasheva2021; Haoming & Wei, Reference Haoming and Wei2024; Lee et al., Reference Lee, Warschauer and Lee2019; and Li & Lei, Reference Li and Lei2022; for grammar, see Rassaei, Reference Rassaei2024; Shintani, Reference Shintani2015; and Shintani et al., Reference Shintani, Li and Ellis2013; for reading, see Chen & Zhao, Reference Chen and Zhao2022; Cheung & Slavin, Reference Cheung and Slavin2012; Graham & Hebert, Reference Graham and Hebert2011; Hall et al., Reference Hall, Roberts, Choo, McCulley, Carroll and Vaughn2017; Maeng, Reference Maeng2014; and Yapp et al., Reference Yapp, De Graaff and van den Bergh2021; for writing, see Kang & Han, Reference Kang and Han2015; Kao & Wible, Reference Kao and Wible2014; Liu & Brown, Reference Liu and Brown2015; and Vuogan & Li, Reference Vuogan and Li2023; for speaking, see Hu et al., Reference Hu, Kuo and Dixon2022; Lee et al., Reference Lee, Jang and Plonsky2015; Mahdi & Al Khateeb, Reference Mahdi and Al Khateeb2019; Saito, Reference Saito2012; Saito & Plonsky, Reference Saito and Plonsky2019; and Sakai & Moorman, Reference Sakai and Moorman2018; for listening, see Dalman & Plonsky, Reference Dalman and Plonsky2022 and Shintani & Wallace, Reference Shintani and Wallace2014; for pragmatics, see Derakhshan & Shakki, Reference Derakhshan and Shakki2021; Plonsky & Zhuang, Reference Plonsky, Zhuang and Taguchi2019; Ren et al., Reference Ren, Li and Lu2023; and Taguchi, Reference Taguchi2015). When designing an ISLA research study, whether a new study or a replication (discussed more in the Replication section), it is imperative that the researcher begin with a thorough review from within ISLA, ensuring that the study is theoretically and empirically grounded, and has the potential for application. This trifold approach provides the greatest likelihood that the study will contribute to the larger goals of ISLA.

Overarching goals and ultimate aims of ISLA

As a research domain, ISLA seeks to impact our understanding of language teaching in three complementary ways: theoretical, empirical, and applied. First, all decisions in empirical research must tie back to theoretical assumptions. Second, ISLA researchers aim to provide a robust body of empirical evidence that increases our understanding of how L2 development occurs in instructed settings. And finally, ISLA research strives to provide research that is useful in application in the real world, especially in pedagogical contexts (see Plonsky, this volume). These foundational considerations will be a touchstone throughout the remainder of the article.

Use of intact classes and/or heterogeneous small participant pools

ISLA studies frequently use intact classes in research designs, which, like all methodological decisions, has benefits as well as drawbacks. Utilizing intact classes increases the ecological validity of the study (i.e., Rogers & Cheung, Reference Rogers and Cheung2021; Sato & Loewen, Reference Sato, Loewen, DeKeyser and Prieto Botana2019; Spada, Reference Spada2005; Spada & Lightbown, Reference Spada and Lightbown2022), is aligned with the overarching goals of ISLA and the heterogeneous reality of instructed contexts, often allows for in-class control of any target items (if applicable), and usually decreases the need for financial compensation, given that participants are enrolled students and often offered extra credit. On the other hand, when using intact classes, time is constrained for a specific study, given the time limitations of a specific instructional term (e.g., a semester, an academic year), which often results in cross-sectional as opposed to longitudinal studies, and smaller sample sizes (made even smaller by attrition if someone is absent during part of the study, or as a result of transient students or students who are frequently out of the classroom due to educational enrichment, interventions, or challenges). While it is possible to continue to recruit student participants who stay enrolled in the program or even afterward, it is not a given that the researchers will continue to have access or participant interest; for once notable exception, see longitudinal studies following students during and after study abroad by Huensch, Mitchell, Tracy-Ventura, McManus and colleagues (Huensch & Tracy-Ventura, Reference Huensch and Tracy-Ventura2017; Huensch et al., Reference Huensch, Tracy-Ventura, Bridges and Cuesta Medina2019; McManus et al., Reference McManus, Mitchell and Tracy-Ventura2021; Mitchell et al., Reference Mitchell, Tracy-Ventura and Huensch2020). To address these considerations, it is important to examine and thoroughly report the nature of the instructional context, including detailed information regarding the participants and their motivation for taking the course, as well as the nature of the instruction and the instructor. In other words, for ISLA studies conducted within an instructional context, the researcher must thoroughly acknowledge that this is the case and not report the study as if it is happening in a laboratory or other highly controlled space. As is necessary for all scientific studies employing quantitative methods and statistical analyses, it is important to proactively and preemptively determine the number of participants needed for each treatment group (when applicable), the power needed for the study (Loewen & Hui, Reference Loewen and Hui2021), and to ensure that all assumptions are met (Hu & Plonsky, Reference Hu and Plonsky2021) before collecting data. In early steps of analysis, data visualization to test for distribution is also advised. This is particularly important for ISLA studies where L2 gains are examined; having both productive and receptive measures of participants’ proficiency levels in a pretest phase, especially with respect to any target item(s). One cannot assume students from the same class are at the same level. Participant IDs must also be preemptively measured and taken into account, to better understand the complexity of an in-tact class. ISLA researchers must decide if it is more important for their study to have in-tact classes participate in treatment groups (when applicable), or if it is more important to have participants of similar variables in specific groups.

Using your own students for ISLA research

Recruiting one’s own students for an ISLA research study is both a unique and important opportunity to be an “insider” in the study and ensures a greater chance of things proceeding as designed. That being said, one’s own students are considered a vulnerable population at risk for coercion according to ethics committees, and usually require additional planning and assurances, such as having another person collect informed consent, not being informed who participated in the study until after final grades are calculated, and not looking at the data for research purposes until after the semester is over. Indeed, Larsson et al. (Reference Larsson, Plonsky, Sterling, Kytö, Yaw and Wood2023) report that “recruiting participants to join a study in a way that makes refusal difficult or uncomfortable” is one of the least reported and most severely questionable research practices in the larger umbrella field of applied linguistics; given the power differential of teachers and students, the field of ISLA is at particular risk for this practice. Additionally, as the researcher will know individual students and bring their own biases on the learners’ performance, interactions, or other L2 variable(s) from prior experiences, having someone else anonymize all data and code as an interrater is particularly important for a study of this kind. In ISLA studies where the researcher knows the participants, one or more common biases may be at play, and are important to preemptively address. For example, a halo effect is the propensity to use an existing positive impression of a participant to influence one’s impression of other areas/behaviors (Behrmann, Reference Behrmann2019; O’Grady, Reference O’Grady2023; Sanrey et al., Reference Sanrey, Bressoux, Lima and Pansu2021). An observer bias can occur when the researcher has influence at the design, implementation, and/or evaluation and interpretation stages of the project; the risk is that the researcher will observe in the data what they are looking for (Derwing & Munro, Reference Derwing and Munro2005). On the flip side of the experience, an observer or Hawthorne effect is the risk that the participant will act differently because they know they are being observed (Greenier & Moodie, Reference Greenier and Moodie2021; Mackey, Reference Mackey, Loewen and Sato2017). When participants act in ways they think the researcher “wants” them to act – which may or may not be congruent with their usual behavior – this is an additional issue referred to as the Rosenthal effect (Tsiplakides & Keramida, Reference Tsiplakides and Keramida2010). These risks can be reduced by having another researcher and/or another stakeholder (see discussion on researcher–practitioner dialogue below) involved in recruiting, collecting, and/or analyzing the data; even better would be to have another person’s unbiased perspective involved at all stages. If an ISLA researcher wants to collect data from their own class during “normal” classroom practices, for example, they may have another researcher come in to get consent for using students’ coursework for data purposes. As mentioned earlier, the researcher/instructor could not have access to the data until the course is over or, alternatively, could only have anonymized data during the course if data interpretation needs to occur during the course. For any ISLA study, it is important to consider the pros and cons of recruiting your own students for each particular treatment group, your own positionality within the research (see King, this volume, for more on positionality), what is needed for a robust study and, even more importantly, weigh the risks and benefits for the students in addition to the potential research knowledge gained (Galloway, Reference Galloway, McKinley and Rose2017).

Individual differences

To thoroughly understand what is occurring in an ISLA context, there would ideally be preemptive and intentional consideration of the IDs of all involved – the learners, the instructor, the researcher(s), and any other interlocutors. For example, in theoretical frameworks such as Sociocultural Theory, this would also include inanimate objects and those not physically present in the space (see Back, Reference Back and Gurzynski-Weiss2020 and Lantolf, Reference Lantolf and Gurzynski-Weiss2020; for additional discussions of IDs within multiple theoretical frameworks, see Gurzynski-Weiss, Reference Gurzynski-Weiss2020). As mentioned earlier regarding ISLA RQ4, IDs are currently considered to be dynamic, changing over both micro and macro timescales alongside and at times because of other factors (see Gurzynski-Weiss, Reference Gurzynski-Weiss2020; Li et al., Reference Li, Hiver, Papi, S., Hiver and Papi2022a; and Serafini & Sanz, Reference Serafini and Sanz2016). While empirical trends have been found regarding specific IDs, especially learner IDs (e.g., L2 motivational selves), there is considerable ID research that has been done using instruments that assume IDs are static (e.g., using a questionnaire at one point in time and categorizing a participant as more or less anxious or motivated). Updating the instrumentation utilized to measure and understand the dynamicity of individual IDs (such as has been proposed for L2 anxiety in Gregersen, Reference Gregersen2020; Gregersen et al., Reference Gregersen, MacIntyre and Meza2014; MacIntyre & McGillivray, Reference MacIntyre and McGillivray2023) is greatly needed to better understand the nature of IDs alone, how they interact with other learner IDs, the IDs of non-learner interlocutors (see Gurzynski-Weiss, Reference Gurzynski-Weiss2017a, Reference Gurzynski-Weiss, Loewen and Sato2017b and Gurzynski-Weiss & Plonsky, Reference Gurzynski-Weiss, Plonsky and Gurzynski-Weiss2017), as well as how IDs and context interact with each other (see Larsen-Freeman, Reference Larsen-Freeman and Gurzynski-Weiss2020). One strategy for making instruments more dynamic is a simple one: make the instrument task- rather than learner-specific. For example, examine learners’ motivation related to a specific task (Jarrett & Gurzynski-Weiss, Reference Jarrett and Gurzynski-Weiss2023; Torres & Serafini, Reference Torres and Serafini2016), in lieu of or – even better – in addition to a more general measure of L2 motivation. Another improved practice is to measure an ID, such as one’s ideal L2 self, for example, at multiple timescales across an L2 program (Serafini, Reference Serafini2020).

Examining the process(es) of L2 learning/development

Given ISLA’s historic adoption from the larger field of SLA, the pretest-treatment-posttest quasi-experimental design was dominant for quite some time. However, this did not allow for data answering the ISLA RQs of how L2 learning occurs in classroom contexts (ISLA RQ1), nor did it allow for an exploration of how contextual (ISLA RQ3) and individual variables (ISLA RQ4) impact ISLA, and what techniques are most effective for learning in instructed contexts (ISLA RQ5). Over the past two decades, more studies have examined both the outcome of L2 learning as well as the process of learning itself (e.g., Kim et al., Reference Kim, Choi, Yun, Kim and Choi2022). Processing elicitation procedures including eye tracking data, neuroimaging data, think aloud protocols, and stimulated recall have been used, alone and at times in conjunction with each other (Grey, Reference Grey2023; Révész & Gurzynski-Weiss, Reference Révész and Gurzynski-Weiss2016; Révész et al., Reference Révész, Michel and Lee2019). These methods, especially when used alongside elicited production data, can provide valuable insight into the nature of learners’ L2 development with respect to a specific competency or skill at a particular moment in time (see Comajoan, Reference Comajoan2019; Kissling & Muthusamy, Reference Kissling and Muthusamy2022; and Liskin-Gasparro, Reference Liskin-Gasparro2000 for examples of these methods used to assess grammar learning; see Nakatsuhara et al., Reference Nakatsuhara, Inoue, Berry and Galaczi2017 for speaking). Moving forward, while research examining the processes of L2 learning and development in instructed contexts has the potential to impact each of the ISLA RQs, the research in this area is particularly well matched to inform ISLA RQ1 – the nature of L2 learning. More nuanced understandings of the relationships between the aforementioned variables of individual, context, and intervention, could be explored with research methods capable of capturing ISLA learning in real time.

Expanding our conceptualization of instructional contexts

While earlier ISLA research studies focused on face-to-face L2 classrooms (Collins & Muñoz, Reference Collins and Muñoz2016; Norris & Ortega, Reference Norris and Ortega2000; Shintani et al., Reference Shintani, Li and Ellis2013), as the field grows, so too does our conceptualization and inclusion of what is an “instructed” context. As both Loewen (Reference Loewen2020a) and Gurzynski-Weiss and Kim (Reference Gurzynski-Weiss and Kim2022b) highlighted, the context qualifies as instructed as long as there is an intentional attempt to learn the L2, and an attempt to facilitate that learning by a more experienced individual, interlocutor, or resource (e.g., a language learning app created by someone with L2 knowledge). Thus, ISLA research is ripe with opportunities to replicate and expand research conducted in face-to-face classroom settings to instructed contexts such as synchronous online, hybrid face-to-face and online, asynchronous online, study abroad, domestic immersion, and the numerous types of language learning apps and platforms. Simultaneously, these newer contexts require novel designs tailored to their uniqueness. Much research is needed expanding our understanding of the nature of ISLA in under-explored instructional contexts (e.g., first-generation students during study abroad in Tracy-Ventura et al., Reference Tracy-Ventura, Washington and Mikheeva2024; refugee populations in Field & Ryan, Reference Field and Ryan2022; Shepperd, Reference Shepperd2022; low-income student populations in K–12 schools in Butler & Le, Reference Butler and V-N2018; Winsler, Reference Winsler, Sekerina, Spradlin and Valian2022; and students with specific learning difficulties in Kormos, Reference Kormos2020; Kormos & Smith, Reference Kormos and Smith2023; Kormos et al., Reference Kormos, Košak-Babuder and Pižorn2019; Košak-Babuder et al., Reference Košak-Babuder, Kormos, Ratajczak and Pižorn2019; Randez & Cornell, Reference Randez and Cornell2023), and researchers must consider how the characteristics of the context (ISLA RQ3) – not as they are assumed to be, but empirically verified – may require modifications (and subsequent validations) to methods, analysis, and interpretation of individual ISLA research designs.

Conducting practitioner-based research

Given that the overarching goal of ISLA research is to maximize learning in instructed contexts, researcher–practitioner collaborations (most often with teachers) are at the heart of ISLA research, and this cooperative research is indeed steadily increasing. Importantly, practitioner-based research does not necessarily mean that all projects should be equally designed, implemented, and analyzed by both researchers and practitioners. Rather, it means that research should begin from a place of potential impact of ISLA theory and prior empirical research, and equally from a place of potential impact for those providing L2 opportunities in instructed contexts (see special issue on the research–practitioner dialogue by Sato & Loewen, Reference Sato and Loewen2022 and Spada & Lightbown, Reference Spada and Lightbown2022 for concrete ideas on how to strengthen the research–practice relationship). Simply put, if the research is not meaningful to practitioners, in its execution and/or in its potential results, in the short or long-term, it is not a worthwhile ISLA study.

This inextricable two-way relationship between ISLA research and practice, where research informs pedagogy which informs research can be strengthened in at least three ways: (a) facilitating collaboration among ISLA researchers, language program directors, and classroom teachers; (b) building in collaborations during teacher training programs; and (c) conducting research on teachers’ awareness of and engagement with research-grounded ideas. With respect to facilitating collaboration between ISLA stakeholders, this can begin with a thorough review of the research topic at hand, moving beyond the fields of SLA and ISLA, and looking at where and how the topic is treated by practitioners. Where are practitioners getting their ideas and resources from? Which resources, readings, online platforms, and/or conferences, and what are the messages being shared? For example, in ISLA research, we have firmly departed from the input hypothesis (Atkinson, Reference Atkinson2011; Mackey, Reference Mackey2020; VanPatten & Williams, Reference VanPatten, Williams, VanPatten and Williams2015), whereas Krashen continues to be a regular invited keynote speaker at recent teacher-focused conferences (AATSP, 2024, FFLA, 2023, and KOTESOL, 2018). Making assumptions that researchers and practitioners are on the proverbial “same page” at best limits the impact one’s research could have. Instead, approaching local schools to form individual partnerships (Gurzynski-Weiss et al., Reference Gurzynski-Weiss, Wray, Coulter-Kern and Bernardo2024; Gurzysnki-Weiss et al., Reference Gurzysnki-Weiss, Wray and Coulter-Kernin press; Lyster, Reference Lyster2019; Martin-Beltran & Peercy, Reference Martin-Beltran and Peercy2014; Seiser & Portfelt, Reference Seiser and Portfelt2024; Tavakoli, Reference Tavakoli2023), or joining and attending local or subject-specific conferences can provide irreplaceable insight into practitioner perspectives and complementary expertise beyond traditional academic venues. Longitudinal research collaborating with practitioners either as authors or paid consultants can be extremely beneficial, as well (Gurzynski-Weiss et al., Reference Gurzynski-Weiss, Wray, Coulter-Kern and Bernardo2024, Reference Gurzysnki-Weiss, Wray and Coulter-Kernin press; Spada & Lightbown, Reference Spada and Lightbown2022).

A second way of encouraging practitioner–researcher collaborations is by building it into a teacher training program, either by conducting a classroom replication study as part of a class (Vásquez & Harvey, Reference Vásquez and Harvey2010), or by requiring an action research project, just to name two examples (see Farrell, Reference Farrell2016 for an overview of action research within TESOL contexts; see the journal Educational Action Research for a collection of theoretical reflections and publications of action research; and see Calvert & Sheen, Reference Calvert and Sheen2015; Kong & Pan, Reference Kong and Pan2023; Schart, Reference Schart, Eckerth and Siekmann2008; and Vaca Torres & Gómez Rodríguez, Reference Vaca Torres and Gómez Rodríguez2017 for examples of published action research studies). In the former example, Vásquez and Harvey (Reference Vásquez and Harvey2010) guided their students in replicating an empirical study (Lyster & Ranta, Reference Lyster and Ranta1997) on oral corrective feedback practices, conducting the replication study from start to finish as a class. In action research, on the other hand, the study intentionally takes place within one’s current classroom setting, with the aim to improve that setting without generalizing the results elsewhere. Calvert and Sheen (Reference Calvert and Sheen2015) worked together in an action research study (that also is a researcher–practitioner collaboration) where Calvert, the teacher, developed, implemented, critically reflected on and modified a language learning task to address the needs of her adult English-learning refugee students; this was accomplished with the collaboration of Sheen as the researcher, as well as the participation of students completing the tasks. A third way to engage practitioners is to design research that empirically examines their unique perspectives. For example, several ISLA studies have examined teacher practitioner notions of task complexity, finding that while teachers are in harmony with each other, citing linguistic structure as their main concern and manipulation when making a task more difficult or easier for particular learner levels (Awwad, Reference Awwad2019; Baralt et al., Reference Baralt, Gurzynski-Weiss, Kim, Sato and Ballinger2016; Hasnain & Halder, Reference Hasnain and Halder2021; Révész & Gurzynski-Weiss, Reference Révész and Gurzynski-Weiss2016; Tavakoli, Reference Tavakoli2009; Zhang & Zhang, Reference Zhang and Zhang2022), these practitioner perspectives do not align with researcher notions of task complexity. More simply put, the current most investigated theoretical framework guiding ISLA research on task complexity (Robinson & Gilabert, Reference Robinson and Gilabert2007) does not align with practitioner perspectives. As ISLA researchers, we have a responsibility to listen and be informed as much as we share knowledge. Ultimately, this will ensure that we are designing and collecting data that has the best chance of advancing our fieldwide RQs, particularly with respect to how variables related to instructed contexts influence L2 learning (ISLA RQ 3) and how IDs play a role (ISLA RQ4).

Replication studies, especially with bi/multilingual learners in diverse contexts

Replication studies (see McManus, this volume, Reference McManus, Gurzysnki-Weiss and Kim2022 and Porte & McManus, Reference Porte and McManus2019), especially replications that utilize mixed methods (Sato, Reference Sato, Gurzynski-Weiss and Kim2022), are increasingly important throughout the larger fields of applied linguistics and SLA, and I would argue equally if not more so in ISLA, given its unique challenges as outlined earlier. Within ISLA, there is a critical need to be inclusive in participant recruitment and therefore the potential benefits of research (ISLA RQ5). Applied linguistics research as a whole has over-relied on WEIRD (Western, educated, industrialized, rich, and democratic) populations (Andringa & Godfroid, Reference Andringa and Godfroid2020), on English language learners (70.3% according to Al-Hoorie et al., Reference Al-Hoorie, Oga-Baldwin, Hiver and Vitta2022), on adolescents and adults in formal instructed settings (Ortega, Reference Ortega2009), and, as one reviewer pointed out, in settings that are relatively well-funded. There is a particular gap in L2 research in low-income K–12 schools (Pufahl & Rhodes, Reference Pufahl and Rhodes2011). Intentionality with participant selection and study design relates to multiple movements including the bi/multilingual (Ortega, Reference Ortega2010, Reference Ortega2013, Reference Ortega and May2014, Reference Ortega2019) and methodological turns in applied linguistics (Byrnes, Reference Byrnes2013), and the ethics “creep” (Haggerty, Reference Haggerty2004) and open science movement in empirical research at large. In addition to choosing participants that represent the diverse reality of instructed L2 contexts, we must be deliberate in our selection of the methods employed and in the ways that we may need to adjust existing instruments. For example, if a well-used instrument conceptualizes a learner ID as static, even if it has been validated and found useful in earlier studies, it would be inappropriate to use the instrument as-is without updating it to align with current theory that conceptualizes IDs as dynamic, and then rerunning measures of validity and reliability. Another area for replication and improvement could be a more robust reexamination of analyses and interpretation. For example, knowing that many studies tend to use multiple regressions even when the assumption of normality is violated (Hu & Plonsky, Reference Hu and Plonsky2021) provides an ideal opportunity to rerun analyses and reinterpret results, ideally from the same dataset for robust comparison. A third area of opportunity is expanding existing ISLA research that relies solely on quantitative or on qualitative data and enhancing the study with complementary data through a mixed methods design (see Creswell & Plano Clark, Reference Creswell and Plano Clark2018; Ghiara, Reference Ghiara2020; Plano Clark, Reference Plano Clark2019; Sato, Reference Sato, Gurzynski-Weiss and Kim2022; and Tashakkori & Teddlie, Reference Tashakkori and Teddlie2003). The most popular mixed-methods approach in ISLA research has been convergent design (intentional complementary and often simultaneous collection of quantitative and qualitative data considered together at the level of analysis, if not before; see Brutt-Griffler & Jang, Reference Brutt-Griffler and Jang2022; Resnik & Dewaele, Reference Resnik and Dewaele2020; and Sánchez-Hernández, Reference Sánchez-Hernández2018), though explanatory sequential (when quantitative data is collected first, then qualitative, with the qualitative techniques serving to better explain the quantitative results, and often adjusted based on the quantitative data gained; Andujar, Reference Andujar2020; Bryfonski & Sanz, Reference Bryfonski and Sanz2018; Mitchell et al., Reference Mitchell, Tracy-Ventura and Huensch2020; Rahimi & Fathi, Reference Rahimi and Fathi2021), or exploratory sequential (when quantitative data is collected first, and is used for its own value as well as informing the design of the quantitative component; Doolan, Reference Doolan2021; Liao & Li, Reference Liao and Li2020; Rahmati et al., Reference Rahmati, Sadeghi and Ghaderi2019; Salvador-García et al., Reference Salvador-García, Capella-Peris, Chiva-Bartoll and Ruiz-Montero2020) mixed methods approaches are also often used. Relying on solely one way of collecting data increases the risk that the method does not fully capture the reality of the phenomenon of study and does not provide an equal opportunity for all participants.

A fourth area for ISLA replication ties back to the emphasis on practitioner–researcher collaboration, and relates to the next trend on justice, equity, diversity, and inclusion (JEDI). Having diverse perspectives reviewing recruitment materials, instruments, quantitative questions and qualitative prompts, methods of analysis, and interpretation, would undoubtedly yield enhanced and complementary data and a more comprehensive understanding of a given study and its contribution to one of the ISLA RQs. ISLA researchers must take care to be aware of their own positionality (King, this volume) and the potential limitations and biases that come with their own experiences, especially when they differ from their targeted participants. If our aim is to advance knowledge as a field, then all stakeholders must be part of and benefit from the ISLA research that is being conducted.

Refining ISLA methods with an eye for ethics and JEDI

In addition to the usual requirements for research with human subjects, there is a movement in applied linguistics to ensure transparent and ethical practices throughout the research process (De Costa et al., Reference De Costa, Randez, Her and Green-Eneix2021a; De Costa et al., Reference De Costa, Sterling, Lee, Li and Rawal2021b; Marsden & Plonsky, Reference Marsden, Plonsky, Gudmestad and Edmonds2018; Plonsky, Reference Plonsky2013). With respect to ethical considerations, as ISLA often utilizes current students, the relationships between student and researcher and student and teacher, as well as with the larger instructional context (ISLA RQ3), need to be taken into consideration and protected, as discussed in Unique Methodological Challenges in ISLA Research above.

ISLA research, like SLA and applied linguistics in general, is undergoing a long overdue reckoning when it comes to JEDI. As mentioned earlier, who we work with, how we recruit and treat our participants, and who can benefit from research findings is extremely important. These changes are happening at the level of linguistic theory, for example the complementary field of Instructed Heritage Language Acquisition, as articulated by Bowles and Torres (Reference Bowles, Torres, Montrul and Polinsky2022), which examines the unique acquisition of a heritage language by participants who grew up listening to and/or speaking and/or culturally identifying with the language. At an empirical level, ISLA researchers could choose to return to studies that have previously excluded heritage learners, for example, or who had marked translanguaging as “incorrect” or “not-target like” and revisit and reanalyze these within today’s framing, assuming the methods utilized are still robust in today’s understanding. Throughout a research project, multicultural and multilingual realities must be taken into account and participant identity, especially the identity of participants who could be otherwise identified based on the characteristics provided in the study, must be protected. These updates also have the potential to impact all of the ISLA RQs, taking care that ISLA is a field that impacts in diverse instructional contexts as it sets out to be.

Conducting open, transparent research that has potential real-world impact and which dialogues with multiple stakeholders at all stages

As in applied linguistics and SLA, research on ISLA is increasingly open and transparent, with efforts to share research materials, data, and findings in open-access venues beyond academic journals. These include individual scholar profiles on ResearchGate, Google Scholar, Academia.edu, as well as faculty or personally maintained websites, and YouTube channels. Fieldwide efforts include those in Table 1.

Table 1. Open science platforms in applied linguistics, SLA, and ISLA

In order to align with the overarching goals and the ultimate aims in theory, research, and practice, ISLA research must be conducted transparently, and in a way that is accessible to all stakeholders, and with the potential for social utility and impact beyond academia. ISLA studies should often involve practitioners and respect complementary expertise, and instruments, data, and findings must be shared as openly as possible.