Abstract
Instructed second language acquisition (ISLA) research has recently attracted more focal attention due to the publications of three books and as the theme of the 2016 Second Language Research Forum (SLRF) conference that celebrated its 35th anniversary in the field of second language acquisition (SLA). Recent definitions (e.g. Loewen, 2015) have underscored the context (instructed vs. naturalistic setting), the ‘mechanisms of learning’ (cognitive processes), and the potential manipulation of these processes or the conditions under which such processes take place by instructional intervention. This article goes a step further to consider the curricular aspect of the ISLA context that exists within the language curriculum, the type of learning that does take place in an instructed environment and should be promoted in the instructed setting, and the pedagogical implications for the instructed L2 environment, given its curricular status (Leow & Cerezo, 2016). To this end, this article (1) provides a critical discussion of the context of ISLA, (2) presents a succinct overview of cognitive processes reported to play a role in the L2 learning process, (3) reports the findings of empirical research on implicit/incidental and explicit/intentional learning, and (4) reports on one recent study that purports to acknowledge these variables. Recommendations for future ISLA research are provided.
Keywords
I Introduction
Instructed second language acquisition (ISLA) has been referenced in the literature of second language acquisition (SLA) for over 35 years, yet it is only recently that three major book-length treatises (Leow, 2015a; Loewen, 2015; Loewen & Sato, 2017), two special issues (Nassaji & Simard, 2010, Nassaji, 2016a), and a timeline of research (Nassaji, 2016b) have appeared. These publications have addressed major theoretical, empirical, and pedagogical perspectives together with a new model of the L2 learning process in ISLA (Leow, 2015a). There has also been a subtle shift in the definition of what comprises ISLA. Early definitions focused on ‘research that concentrates on how [author’s italics] a theoretically and empirically based field of academic inquiry that aims to understand how the systematic manipulation of the mechanisms of learning and/or the conditions under which they occur enable or facilitate the development and acquisition of a language other than one’s own. (Loewen, 2015, p. 2; author’s italics)
These definitions underscore (1) the instructed setting, (2) the focus on the ‘mechanisms of learning’ (cognitive processes) employed in this instructed setting, that is, how L2 learners process L2 data in this setting, and (3) the potential manipulation of these processes by instructional intervention with the assumption that superior or faster L2 development will result. According to Leow and Cerezo (2016), there are also three additional important features that need to be associated with any definition of ISLA, namely, (1) the curricular aspect of ISLA (that is, its place within the language curriculum), (2) the type of learning that takes place in an instructed environment and should be promoted in this setting, and (3) the pedagogical implications for the instructed L2 environment given its curricular status. This article (1) provides a critical discussion of the context of ISLA, (2) presents an overview of cognitive processes reported to play a role in the L2 learning process, (3) reports the findings of empirical research on implicit/incidental and explicit/intentional learning, (4) reports on one recent study that purports to acknowledge these issues, and (5) provides recommendations for future ISLA research.
II The context of ISLA
The ‘I’ in ISLA clearly underscores that ISLA research is situated in some formal or instructed L2 environment (classroom) in which instruction or formal exposure to the L2 is either provided by an instructor or via hybrid/blended or online learning courses (Allen & Seaman, 2013). The L2 is a naturally-occurring language and the so-called traditional four skills of listening, reading, writing, and speaking or a combination thereof are promoted within a relatively short period of time. In this setting, there are many variables in perennial play that include learner (Li, 2017) and teacher (Gurzynski-Weiss, 2017) characteristics. More importantly, ISLA is also situated within a curriculum that has specific goals, curricular information, and learning outcomes. Students follow a syllabus, a prescribed textbook, and homework is typically based on the content of this textbook. Very importantly, success in such classes is dependent upon a grade that usually ranges from an A to an F. Consequently, ISLA research that seeks to provide pedagogical implications needs to consider the usefulness of its findings in relation to the curricular learning outcomes of the instructed setting.
ISLA research that seeks to probe into learner cognition, learning conditions, and their potential manipulations, then, needs to focus on the identification and roles of the cognitive processes employed by L2 learners in the typical instructed setting. To this end, the next section provides a succinct report on cognitive processes postulated to play a role in the L2 learning process.
III The role of cognitive processes in theoretical underpinnings of ISLA
A cursory review of theoretical underpinnings 1 cited in the cognitive ISLA literature to account for L2 development reveals quite an impressive number of underpinnings that are primarily grounded within two perspectives: (1) psychology-based, such as McLaughlin’s (1987) cognitive theory, Schmidt’s (1990 and elsewhere) noticing hypothesis, Tomlin and Villa’s (1994) model of input processing in SLA, Robinson’s (1995) model of the relationship between attention and memory, DeKeyser’s (2007) skill acquisition theory, N.C. Ellis’s (2007) associative-cognitive framework, VanPatten’s (2007) input processing model/theory, Truscott and Sharwood-Smith’s (2011) MOGUL (Modular Online Growth and Use of Language), and Leow’s (2015a) model of the L2 learning process in ISLA, or (2) interaction/psychology-based, such as Gass’s (1997, expanded in Gass and Selinker, 2008) model of second language acquisition and Swain’s (2005) output hypothesis. In addition, it is noted that these theoretical underpinnings can be situated along stages postulated to occur along the L2 learning process, viewed from both a process and product perspective, visually seen in Table 1.
Stages of the L2 learning process in ISLA (Leow, 2015a, 2015b).
All these theoretical underpinnings share in some degree several major cognitive processes (attention, awareness, activation of prior L1/L2 knowledge, working memory and variables such as levels of awareness and levels or depth of processing) assumed to play important roles during the L2 learning process. Indeed, these cognitive processes, postulated to account for the preliminary exposure to L2 input and learners’ eventual output, lend credence to recent definitions that have specifically focused on the role and manipulation of such cognitive processes in ISLA. These cognitive processes and variables are provided in Table 2.
A synopsis of the cognitive processes and variables postulated to play important roles in the second language (L2) learning process.
Note. Items in parentheses: not clearly stated but gleaned from postulation.
However, there is a strong debate regarding how L2 input needs to be processed, that is, what type of learning (explicit/intentional vs. implicit/incidental) is important for subsequent learning or internalization of the L2 input (e.g. Hulstijn, 2013; N.C. Ellis, 2015; Leow, 2015a; for a review, see also Nassaji, 2017). This debate has led to several studies attempting to probe empirically the role (un)awareness plays in the L2 learning process. The construct ‘awareness’ has been defined, for example, as ‘a particular state of mind in which an individual has undergone a specific subjective experience of some cognitive content or external stimulus’ (Tomlin & Villa, 1994, p. 193) or ‘refers to a state of mind in which one has become cognizant of the regularities underlying the data’ (Schacter, 1989, p. 577).
While many theoretical underpinnings appear to postulate some important role for awareness at the input to intake processing stage (Schmidt, Robinson, Gass, VanPatten), Tomlin and Villa, Truscott and Sharwood-Smith, and Leow do not. While Tomlin and Villa reject any important role for awareness at this stage, Truscott and Sharwood-Smith and Leow hinge this important role on some high level of activation or depth of processing, respectively, for further processing to take place beyond intake. For example, Leow subsumes the role of awareness within the notion of depth of processing at the intake processing stage of the L2 learning process. Depth of processing is defined as ‘the relative amount of cognitive effort, level of analysis, elaboration of intake together with the usage of prior knowledge, hypothesis testing and rule formation employed in decoding and encoding some grammatical or lexical item in the input’ (Leow, 2015a, p. 204). In other words, how L2 learners process L2 data beyond intake will be indicative of the type of learning (explicit vs. implicit) that takes place in ISLA. Implicit learning is marked by a very low depth of processing without much cognitive or mental effort employed to process the L2 data. On the other hand, any processing that exceeds this low depth of processing has the potential to lead to some level of awareness to the extent that as the depth of processing increases to include hypothesis testing and rule formulation, so too does the potential level of awareness increase: from awareness at the level of noticing to awareness at the level of reporting to awareness at the level of understanding (see Leow, 2001a, 2015a; for a report of this correlation between depth of processing and levels of awareness, see Leow, 2012).
It is well documented in the (I)SLA literature that the roles of explicit learning (or learning with awareness) and intentional learning (learners are specifically requested to learn target L2 data) have positive benefits on L2 development (see, for example, Cerezo, Caras, & Leow, 2016; de la Fuente, 2016; Hsieh, Moreno & Leow, 2016; Leow, 1997, 1998a, 1998b, 2000; Medina, 2016; Rosa & Leow, 2004; Rosa & O’Neill, 1999; Sachs & Suh, 2007 for explicit learning and Barcroft, 2009; Hamrick & Rebuschat, 2014; Hulstijn, 1992; Kachinske, Osthus, Solovyeva, & Long, 2015 for intentional learning, see Paribakht & Wesche, 1996). More importantly, several studies employing concurrent data elicitation procedures to address the role of cognitive processing and processes have provided empirical evidence supporting a correlation between higher depth of processing or level of awareness and amount of L2 development (e.g. Adrada Rafael, 2017; Bird, 2012; Bowles, 2003; Calderón, 2013; Hsieh, Moreno, & Leow, 2016; Leow, 1997, 2001b; Rosa & Leow, 2004; Rosa & O’Neill, 1999; Sachs & Suh, 2007; see also studies on Laufer and Hulstijn’s (2001) Involvement Load Hypothesis, e.g. Hulstijn & Laufer, 2001; Keating, 2008; Kim, 2008; Rott, 2005). However, to address the crucial question regarding the type of learning that is beneficial to L2 development in ISLA, the next section provides a concise report of the literature on the roles of implicit/incidental vs. explicit/intentional learning in ISLA with a special focus on the role of awareness or lack thereof in L2 learning.
IV The role of implicit/incidental vs. explicit/intentional learning in ISLA
Whether one can learn a foreign or second language implicitly, typically defined as learning without awareness or incidentally, 2 typically defined as the absence of any deliberate intention before exposure to the L2 data to learn target L2 information in such input, has been an issue that has not only permeated several studies in the non-SLA field for decades but also in the current field of (I)SLA. Indeed, the empirical origins of implicit and incidental learning date back to the beginning of the 20th century, and began in psychology-based studies (for example, for incidental learning, see Jenkins, 1933; for implicit learning, see Thorndike & Rock, 1934).
Early notions of implicit and incidental learning, from both a vocabulary and grammatical perspective, appear to have a close connection to Krashen’s (1982) Monitor Model in the ISLA literature that distinguished between the constructs ‘acquisition’ and ‘learning’ and, in the field of cognitive psychology, Reber’s seminal 1967 and other studies (e.g. Reber, Kassim, Lewis & Cantor, 1980) that investigated these types of learning employing artificial or finite-state grammars that generate meaningless letter strings. According to Krashen, the acquisition process is subconscious and ‘effortless’, given that the learner processes the language with minimal amount of cognitive or mental effort. Acquisition is also described by Krashen as ‘… implicit learning, informal learning, and natural learning. In non-technical language, acquisition is picking up a language’ (p. 10), which appears to suggest, as pointed out in Leow (2015a), that acquisition, incidental learning, and implicit learning all share two important features: an absence of awareness and a low depth of metalinguistic processing during the learning process. The conflation between these types of learning is seen in the direct association between the acquisition process and incidental learning: ‘Thus, the acquisition process is identical to what had been termed ‘incidental learning’ (R. Ellis, 1994, p. 212). Reber (1976) defines implicit learning as a ‘process whereby a subject becomes sensitive to the structure inherent in a complex array by developing (implicitly) a conceptual model which reflects the structure to some degree’ (p. 88).
The role of awareness or lack thereof began to be addressed in several permutations of learning conditions that appeared to conflate specifically the following two types of learning, namely, implicit and incidental in opposition to explicit or intentional learning. The typical design comprised learning conditions in which some participants were provided with either grammatical information on or instructions to learn the experimental data (explicit or intentional) while others were simply exposed to the target information (implicit or incidental). However, the construct of awareness was operationalized from two different approaches, namely, at the reconstruction stage (non-concurrent) or during the construction or encoding stage (concurrent). 3 At the non-concurrent stage, awareness was operationalized and measured after the experimental exposure to the L2 data via offline verbal reports (e.g. Chan & Leung, 2014; Faretta-Stutenberg & Morgan-Short, 2011; Grey, Williams, & Rebuschat, 2014; Hamrick & Rebuschat, 2014; Leung & Williams, 2011, 2012, 2014; Rebuschat & Williams, 2012; Rebuschat, Hamrick, Sachs, Riestenberg, & Ziegler, 2013, 2015; Rogers, Révész, & Rebuschat, 2016; Williams, 2005). Other studies opted to employ a concurrent data elicitation procedure (e.g. Adrada-Rafael, 2017; Calderón, 2013; de la Fuente, 2016; Hama & Leow, 2010; Hsieh, Moreno, & Leow, 2016; Leow, 2001a, 2001b; Leow, 1997, 2000; Medina, 2016; Rosa & Leow, 2004; Rosa & O’Neill, 1999; Sachs & Suh, 2007) in an effort to establish the role of awareness or lack thereof during the L2 learning process or how participants were processing the L2 data. This procedure elicited non-metacognitive think aloud protocols in which participants were requested to say aloud whatever they were thinking as they performed the experimental task without the need to provide any explanation for their thoughts. Crucially, coded protocols revealed correlations between levels of awareness and levels of processing in that more cognitive effort put into processing L2 data were associated with higher levels of awareness (for further discussion, see Leow, 2012). Viewed, then, from a processing perspective (Leow, 2015b), if the baseline for the presence or absence of awareness were associated with a very low level of cognitive effort, low depth of processing, speed, automaticity, absence of focal attention, use of minimal attentional resources and so on (descriptors associated with the acquisition or implicit learning process), then any deviation from these processing descriptors would be associated with the presence of some level of awareness.
There are limitations associated with both approaches to operationalizing awareness to address the benefits of type of learning. The major limitation related to gathering awareness data at the concurrent construction stage is the potential role of reactivity. Reactivity addresses ‘whether thinking aloud could have affected participants’ primary cognitive processes while engaging with the L2 or even added an additional processing load or a secondary task on participants, which would not reflect a pure measure of their thoughts’ (Leow, 2015a, p. 142). At the non-concurrent reconstruction stage, limitations include the inability to (1) methodologically establish participants’ behavior during the experimental phase of the study (did they follow instructions regarding the experimental procedure?), (2) ascertain whether the offline performance reflects accurately the learning behavior of each experimental learning condition (did some participants in the implicit or incidental learning condition become aware of the target L2 information?), and, more specifically, (3) gather data on how participants actually processed (low or higher depth of processing) the target information. In addition, it is noted that the relatively popular use of a semi-artificial language or artificial lexicon as the experimental L2 input in many of the incidental or implicit learning condition studies may not reflect the processing of natural languages (e.g. Chen et al., 2011; Faretta-Stutenberg & Morgan-Short, 2011; Graham & Williams, 2016; Grey et al., 2016; Hama & Leow, 2010; Hamrick & Rebuschat, 2014; Kachinske et al., 2015; Leung & Williams, 2011, 2014; Marsden, Williams, & Liu, 2013; Rebuschat et al., 2013, 2015; Rogers et al., 2016; Tagarelli, Ruiz, Moreno Vega, & Rebuschat, 2016; Williams, 2005).
V Incidental vs. intentional learning: What the research reveals
Based mostly on chance performance and semi-artificial language data, the majority of the extant studies appear to provide some empirical evidence that adult L2 learners may incidentally learn aspects of non-native syntax or morphosyntax while processing the language input for meaning and without any instruction to search for or learn a rule (e.g. Grey et al., 2014; Hamrick, 2014; Kachinske et al., 2015; Rebuschat & Williams, 2012; Robinson, 1995; Rogers et al., 2016) and even after a delay of two weeks (e.g. Grey et al., 2014). This type of incidental learning also appears to lead to both implicit and explicit knowledge (e.g. Hamrick & Rebuschat, 2014; Rebuschat & Williams, 2012; Rebuschat et al., 2013, 2015; Rogers et al., 2016), as measured primarily on grammaticality judgment tests, indicating that different types of processing (implicit and/or explicit) are not an unusual occurrence in such a learning condition. Studies on incidental vocabulary learning from reading have also reported some learning of new words (e.g. Day, Omura, & Hiramatsu, 1991; Godfroid, Boers, & Housen, 2013; Krashen, 1989; Pitts, White, & Krashen, 1989), albeit with low percentages. At the same time, it is noted that the studies that compared incidental and intentional learning conditions (e.g. Barcroft, 2009; Denhovska, Serratrice, & Payne, 2016; Hamrick & Rebuschat, 2014; Kachinske et al., 2015) all reported that more learning occurred in the intentional learning conditions. Mean percentages obtained by the incidental groups on the chance tests usually fell between a range of 13% and 64% while the intentional groups were substantially above this range, falling in the 45%–73% range.
VI Implicit vs. explicit learning: What the research reveals
While several studies (e.g. Chan & Leung, 2014; Chen et al., 2011; Graham & Williams, 2016; Leung & Williams, 2011, 2012, 2014; Marsden et al., 2013; Williams, 2005) have reported empirical evidence for implicit learning of form–meaning connections or phonology (via chance performance, reaction times data, and/or non-concurrent operationalization of the construct awareness), other studies, employing both concurrent and non-concurrent data-elicitation procedures, have failed to provide such evidence (Chen et al., 2011; Faretta-Stutenberg & Morgan-Short, 2011; Hama & Leow, 2010; Leow, 2000). However, similar to the incidental versus intentional learning condition comparisons, studies that compared aware and unaware performances also reported substantially more learning in the aware than in the unaware group (Kachinske et al., 2015; Leow, 2000; Rebuschat & Williams, 2012; Rebuschat et al., 2013; Williams, 2005). Mean percentages obtained by the unaware groups on the chance tests usually fell between a range of 49% and 64% while the aware groups were substantially above this range, falling in the 60%–91% range. The single study (Leow, 2000) that employed a naturally occurring language (Spanish) reported gain scores of 55% and 44.4% (aware group) versus 5% and 1.8% (unaware group) on a recognition and controlled written production assessment task, respectively.
In ISLA research, there are a multitude of different approaches, methods, procedures, and techniques (e.g. focus on form, processing instruction, computerized instruction or exposure, textual enhancement, type of feedback etc.) that have been investigated to ascertain their success in promoting L2 learning in the instructed setting (for a review of recent studies in these different areas, see Nassaji 2015). To address the overall effect of type of instruction and indirectly type of learning, four meta-analyses (Kang, Sok, & Han, 2018, Goo, Granena, Yilmaz, & Novella, 2015; Norris & Ortega, 2000; Spada & Tomita, 2010) have lumped many of these different instructional interventions into two types of instruction: explicit and implicit. An instructional treatment, following DeKeyser (1995) and Norris and Ortega (2000), was coded explicit if rule explanation formed part of the instruction (deductive instruction) or if participants’ attention was drawn to specific target items in the L2 input and they were requested to arrive at some metalinguistic rule on their own (explicit inductive instruction). Implicit instruction represented the absence of any rule explanation or any request to arrive at the underlying rule. Three of the meta-analyses supported the use of explicit instruction over implicit instruction: Goo et al., Norris and Ortega, and Spada and Tomita all reported larger effect sizes obtained by explicit over implicit instruction on the immediate posttest. On the other hand, a more recent meta-analysis (Kang et al.) reported relatively similar effect sizes on the immediate posttest but larger effect sizes for implicit instruction on the delayed posttest when compared to explicit instruction.
What do these meta-analyses reveal in relation to type of learning (implicit or explicit)? While it may be argued that in the explicit instructional conditions some degree of explicit learning would have taken place, what may account for the superiority of the implicit over the explicit instructional intervention reported in Kang et al.? Perhaps in the more recent and increasing number of studies coded as implicit instruction, participants were investing more cognitive effort during the instructional phase. This explanation may be supported by several so-called implicit learning conditions (e.g. Morgan-Short & Bowden, 2006; Rassaei, 2014) that included the provision of salient feedback, which may, in turn, have promoted deeper processing of the target L2. In other words, type of instruction (as in deductive instruction) or instructional intervention (as in, for example, textual enhancement) is an external manipulation of the L2 while type of learning is an internal process, which needs to be empirically established.
To summarize, in addition to the beneficial effects of explicit and intentional learning on L2 development (see also the reports of the meta-analyses cited above for type of instruction), research on implicit learning and incidental learning conditions appears to indicate that adult L2 learners may also learn L2 data either implicitly or incidentally in experimental learning conditions. However, whether this type of learning can be extrapolated to the instructed setting, when viewed from both a contextual and curricular perspective, warrants some caution. It is noted that in these studies (1) the L2 input provided was typically semi-artificial language data, (2) learning was assessed in many instances by a chance test or a receptive assessment task, (3) participants were, unnaturally, exposed to a relatively large amount of exemplars, yet the overall performances were low, and (4) participants in the intentional or explicit learning conditions, which is underscored by the very type of explicit processing that typically occurs in any instructed environment, outperformed the implicit and incidental learning conditions. In addition, the face-to-face (FTF) instructed setting is not standard given the multiple variables, such as individual differences, absenteeism, language requirement versus minor/major, teachers’ and students’ characteristics, different curricula etc., that clearly play many differential roles.
Given the need to acknowledge (1) the context (instructed setting) of L2 learning, (2) the curricular goal of promoting robust learning in this specific context, (3) the beneficial role cognitive processes play in the process and product of learning in ISLA, and (4) the revelations of the pertinent research on type of L2 learning, the next section provides a report of one study that purports to adhere to these four variables.
VII Cerezo, Caras, & Leow (2016)
Situated within Leow’s (2015a) model of the L2 learning process in ISLA, Cerezo et al. (2016) employed a video game that implemented ‘guided induction’ to successfully instruct complex grammar (the Spanish gustar structure) online. Guided induction is an instructional approach in which teachers help learners co-construct grammar rules by directing their attention to relevant aspects in the input, asking guiding questions, or both. 4 More specifically, the videogame was carefully designed to promote deeper processing (explicit learning) of the target structure as participants played the game by incorporating the following three major features of an e-tutor 5 (Leow, 2015a, pp. 255–256): (1) ‘task-essentialness’ (Loschky & Bley-Vroman, 1993), that is, participants need to minimally pay attention to the targeted items in the task in order to successfully complete the task, (2) concurrent implicit feedback (to confirm or disconfirm previous hypotheses or rule formulations facilitated by task-essentialness (Leow, 1997; Rosa & Leow, 2004), and (3) prompts (e.g. Lyster & Izquierdo, 2009) that encourage deeper processing (e.g. hypothesis formulation or testing) as participants interact with the L2 data.
Seventy English-speaking learners of beginning Spanish were randomly divided by section into three groups: Guided Induction (GI), Deductive Instruction (DI), and Control. The GI group played the gustar maze video game that was designed to promote a high depth of processing to include hypothesis formulation, activation of recent prior knowledge, metacognition, and awareness at the level of understanding (Schmidt, 1990).
The DI group attended a typical deductive classroom lesson. A teacher asked for translations of the same 20 questions posed to the GI participants, wrote the correct translations on the blackboard, and explained the rules for each constituent, repeating this process for the same 20 exemplars in the videogame. Like typical classroom lessons, participants were also allowed to ask for elaboration during this session. Finally, participants in the control group simply performed the assessment tasks without any formal exposure to the targeted structure.
The results on two controlled production tasks (written and oral sentence translation) immediately after the treatment and 2 weeks later revealed that while both instruction groups (GI and DI) improved significantly across time, outperforming the control group, GI achieved higher learning outcomes on most productive posttests and experienced greater retention. More specifically, while the gain scores on the immediate posttests for both GI and DI groups indicated relatively robust learning (GI: 83%, 91.3% and DI: 63.2%, 60.2% for the oral and written production assessment tasks, respectively), only the GI group maintained such learning after two weeks on the delayed posttests (GI: 72.6%, 81.6% vs. DI: 38.2%, 39.7% for the oral and written production assessment tasks, respectively). Based on the participants’ robust learning outcomes, Cerezo et al. concluded that the video game-based instruction could replace the FTF instruction, which would allow teachers to migrate complex L2 material online to free up classroom time for communicative practice.
As can be observed, Cerezo et al. (2016) viewed instructional intervention from a curricular perspective (Leow & Cerezo, 2016). This approach places a premium on the need to acknowledge the impoverished amount of exposure to which L2 learners are exposed, the curricular learning outcomes, and the challenge to promote robust learning within a relatively short period of time. The study, then, was designed to investigate the potential to migrate complex L2 material online to free up classroom time for communicative instruction (FTF vs. online) while promoting robust learning that is above the passing grade. To promote explicit learning, the study also acknowledged the increasing use of technology and hybrid curricula in the instructed setting (Allen & Seaman, 2013), the affordances of the computer-assisted language learning (CALL) platform, and previous research on the usefulness of e-tutors to experimentally manipulate learner cognitive processes.
The performance of the prototypical explicit instructional intervention (DI) is not unlike that reported in many other studies that employed some grammatical information (e.g. Doughty, 1991; Erlam, 2003). The result does support the conclusion provided by several meta-analyses that explicit instructional intervention does lead to statistically higher performances on the immediate posttest. However, whether retention is also positively affected remains to be robustly supported given that a typical (significant) decrease in performance on the delayed posttest is usually reported (e.g. Morgan-Short & Bowden, 2006; Rosa & Leow, 2004). What is noteworthy in the Cerezo et al.’s study is the retention ability of the GI that may have been associated with the high degree of cognitive engagement, as revealed in the think aloud protocols, during the experimental phase (for further elaboration on the potential role of depth of processing, see Leow, 2015a).
VIII Future directions for ISLA seeking to inform practice in the L2 classroom
The context (instructional setting), type of learning (explicit), and curricular issues are three variables that need to be seriously considered by ISLA research aimed at addressing and contributing to (ideally robust) L2 development in any instructional setting. First of all, from a contextual perspective, as pointed out in Leow and Zamora (2017), it is not uncommon for some researchers to ground their theoretical underpinnings in child acquisition, for example, statistical learning (Marsden et al., 2013), sequence learning (Williams, 2010), or Krashen’s (1982) Monitor Model. For Leow and Zamora, it is important to consider ‘(1) the huge disparity between L1 acquisition and L2 learning in regard to amount and type of exposure to and interaction with the L1 or L2 data and (2) the depth of processing associated with type of learning’ (p. 44).
Second, the typical ISLA setting is designed to promote more explicit and intentional learning than implicit and incidental learning and acquisition. This setting does not negate or preclude any instance(s) of incidental or implicit learning taking place but, as Leow (2015a, p. 244) cautions: this kind of processing depends heavily on many factors that include the provision of large amounts of exemplars in meaningful contexts and quite a long period of time to process, internalize the exemplars, and have the knowledge available for subsequent usage.
However, the findings of ISLA research arguably support the promotion of intentional and explicit learning over incidental and implicit learning in the instructed setting, irrespective of type of instruction. Consequently, further investigation and promotion of cognitive processes associated with great depths of processing, high levels of awareness (hypothesis testing and rule formulation), and activation of both recently learned and prior knowledge are clearly warranted for all types of L2 linguistic and lexical items at all proficiency levels.
Third, ISLA is situated importantly within a language curriculum with its outcome goals, textbook, syllabi, limited exposure, tests, grades, and so on. ISLA research that seeks to inform practice in the instructed setting needs to be grounded in robust learning outcomes or gain scores that would qualify for minimally a passing grade in the instructed setting and not based solely on statistical data such as p values and effect sizes. Such ISLA research that produces large effect sizes but low gain scores that fall way below a passing grade can be used as the foundation for further fine-tuning and development that may ultimately lead to more robust learning outcomes, which can then be pedagogically extrapolated to the instructed setting.
Probing deeper into the roles of incidental/implicit learning in adult L2 learning is of clear theoretical value to the field of SLA. However, viewed from processing, contextual, and curricular perspectives together with the empirical findings of demonstrated superiority of intentional and explicit learning over incidental and implicit learning, ISLA research may better inform teachers, language curricula, and teaching methodology by focusing on the potential roles either intentional or explicit learning (see also N. Ellis, 2015; Leow, 2015a) may play in promoting more robust learning in this setting. A strong ISLA research agenda that seeks to inform practice, then, may be to continue probing deeper into the cognitive processes employed by L2 learners as they interact with or are exposed to the L2 across different modalities, different instructional interventions, types of tasks, linguistic items, language levels, and so on. A better understanding of these processes can contribute to the creation of theoretically-driven and empirically-supported pedagogical tasks or activities (instructional interventions) that are designed to encourage active use of students’ mechanisms of learning while performing such tasks or activities. To this end, ISLA research can be viewed from minimally two perspectives: one that is centered in the traditional setting (e.g. in a FTF format with the teacher and students) with a focus on communicative interaction and practice with the L2 and the other in the online medium that provides individual practice via tasks or videogames that are carefully designed to promote cognitive engagement and types of processing that are theoretically-driven and empirically supported to produce robust learning outcomes.
In closing, the role of ISLA research is arguably to contribute to a better understanding of the many variables that contribute to (robust) language learning and teaching in a relatively impoverished setting when compared to the naturalistic setting in which acquisition occurs. This article supports the recent definition of ISLA (Loewen, 2015) that underscores the important role of cognitive processes in the L2 learning process and how to manipulate such mechanisms via instructional intervention to promote L2 development. To this end, ISLA researchers are urged to probe deeper into the role of cognitive processes and to acknowledge (1) the broader picture in which the ISLA context lies, that is, within the language curriculum, (2) the type of learning (explicit) that does take place in this context and should be promoted in the instructed setting, and (3) the curricular value of pedagogical implications, ideally robust, for the instructed L2 environment.
Footnotes
Conflict of Interest
Also, if this study is part of a larger study or if you have used the same data in whole or in part in other papers, both already published or under review please state where the paper is published and describe clearly and in as much detail as you think necessary where the similarities and differences are and how the current manuscript makes a different and distinct contribution to the field.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
