by Rod Ellis
There is also a bigger issue. The general advantage for explicit instruction is evident in studies where the instruction is directed at learning specific linguistic features—i.e. a particular grammatical structure or a set of vocabulary items. But, as we saw in Chapter 9, there is more to learning a language than acquiring new linguistic features and items. Learners need to also develop interactional competence (i.e. the ability to utilize their linguistic resources in communicative routines in relevant communities of practice). Thus, to establish the relative contributions of explicit and implicit instruction, it is necessary to investigate not just which type of instruction is superior in helping learners acquire isolated linguistic elements but also the relative effects of the two types of instruction on L2 development more generally and, in particular, on learners’ ability to engage confidently and easily in the use of the L2. Arguably, implicit instruction, which adopts a more holistic view of L2 proficiency and seeks to develop not just linguistic but also interactional competence, is better equipped to foster L2 development overall.
Few studies have taken this broader approach to comparing explicit and implicit instruction. One that has done so is Shintani’s study of young beginner learners of English (Shintani and Ellis, 2010; Shintani 2012; Shintani 2015), which was referred to earlier in this chapter and also in Chapter 10. Here is a summary:
This study compared the relative effects of implicit instruction consisting of input-based tasks and explicit instruction consisting of presentation-practice-production (PPP) on the acquisition of a set of new words. The results indicated that both types of instruction were effective, but that overall the implicit instruction was superior.
The study also reported that the implicit instruction resulted in superior incidental acquisition of a grammatical feature (plural -s). This feature was not explicitly taught in either the task-based instruction or in the PPP.
The study also undertook a micro-genetic analysis of the interactions that occurred in the two types of instruction to examine how affordances for the use of the target items (both lexical and grammatical) arose. It showed how opportunities for learners to initiate discourse and to negotiate for both meaning and form were common in the input-based lessons and how—in contrast—the explicit instruction largely constrained learners’ contributions to responding in initiate-respond-follow up (IRF) exchanges.
If we want to have a clearer idea of the relative effects of implicit and explicit instruction we need more studies like Shintani’s including those involving adult learners who are, perhaps, better equipped to take advantage of explicit instruction. We need studies that measure the effects of implicit and explicit instruction on both planned and unplanned linguistic features and that also document the kinds of interactions that occur and how these, to a greater or lesser extent, foster interactional competence.
Conclusion
The goal of implicit instruction is the same as that of explicit instruction, namely the development of the kind of knowledge needed to engage effectively in communication (especially oral communication). However, it differs from explicit instruction in that it does not assume that explicit knowledge is the starting point. Instead it assumes that classroom learners, irrespective of their age, are able to acquire implicit knowledge by participating in activities that foster its incidental acquisition.
Such a position is clearly in opposition to the strong-interface position, but can be seen as compatible with the non-interface position—i.e. explicit knowledge and implicit knowledge are cognitively differentiated and involve different learning processes. Implicit instruction assumes that learners will acquire the L2 naturally and easily through participating in instructional activities that provide them with comprehensible input, opportunities for pushed output, and also attract attention to form. Thus—like explicit instruction—it emphasizes the importance of conscious attention to form. The difference is that implicit instruction draws on theories that claim that attention to form needs to be achieved while learners are trying to communicate. In this way, as the weak-interface position proposes, explicit knowledge can help to activate the processes involved in acquiring implicit knowledge.
There can be no doubt that much of language learning occurs incidentally. As Schmitt (2008) commented, there are limits to how often a teacher can ensure explicit exposure to new words and thus ‘many of the meetings that learners need to consolidate and enhance their knowledge of lexical items must come from the extensive exposure generated by the meaning-focused strand, from which incidental learning can occur’. (p. 246).
The same is true for grammatical items and, perhaps, for phonological features too. It is, however, not just a question of ‘extensive exposure’. As this chapter has attempted to show, it is how opportunities for incidental acquisition can be engineered by the kinds of tasks used and how they are implemented. The theories I have examined may not provide conclusive evidence about how this is best achieved, but they illuminate the factors that are involved and provide a basis for research-informed implicit language pedagogy.
Finally, there is no need for teachers to make a choice between explicit and implicit forms of instruction. Chapter 10 provided clear evidence that explicit instruction is effective. This chapter has shown that implicit instruction is also effective. A language curriculum that includes both explicit and implicit instructional components is perhaps most likely to ensure that language pedagogy is efficient and effective for ensuring balanced L2 development.
Notes
1 A wide range of specific measures have been used to investigate complexity, accuracy, and fluency. See Ellis and Barkhuizen (2005) and Housen, Kuiken, and Vedder 2012) for details of the measures that have been used.
2 Of course, even in implicit instruction that is entirely meaning centred, individual learners may elect to attend to engage in intentional language learning. However, meaning-centred instruction is not designed to encourage this.
3 It was not possible to measure receptive knowledge of copula be as this grammatical structure conveys no meaning—i.e. it is redundant—and thus does not have to be processed for comprehension to take place.
4 Both Coughlin and Duff’s and Seedhouse’s point about the lack of correspondence between the task-as-workplan and the activity that arises when it is performed was made with reference to output-based tasks. Arguably, the same point is relevant to input-based tasks. Shintani (2012), for example, documents in detail how the performance of the input-based tasks in her study varied considerably when they were repeated nine times.
5 However, Tavakoli and Foster (2011) suggested that the increase in complexity may have simply reflected the learners’ use of co-ordinators such as ‘while’ and ‘during’, rather than the complexity involving greater subordination.
6 A number of studies have examined what learners do when they engage in pre-task planning. See for example Pang and Skehan (2014).
7 Wang (2014) investigated two pressured online-planning conditions. In one, they watched a video whilst speaking, while in the other, they watched the video first and then spoke. The effects for online planning were only evident in the first of these conditions. In fact, this condition corresponded closely to the condition in Yuan and Ellis (2003) as the learners in this study were allowed to look at the picture story for a short time before they spoke and also while speaking. This suggests that the online planning condition works in conjunction with another variable—i.e. here-and-now vs there-and-then. In other words, unpressured online planning benefits accuracy when there is also visual support.
8 Samuda’s (2001) study can be seen as an example of Integrated Instruction, which I discussed in Chapter 10. However, Samuda intervened only once and only briefly to explain the use of the modal verbs in a manner I consider more compatible with task-based teaching.
12
Understanding and applying second language acquisition
Introduction
SLA is still quite young as a field of study—much younger
than well-established disciplines such as psychology or sociology. While there has always been an interest how people acquire second languages, the systematic study of L2 acquisition did not start until the 1960s when Chomsky’s (1959) critique of Skinner’s Verbal Behavior liberated researchers from what Larsen-Freeman (2007) termed the ‘bondage of behaviorism’, making it possible to investigate language learning as a cognitive enterprise. Since then, SLA has developed exponentially and, arguably now manifests all the characteristics of an academic discipline (see Table 12.1).
Characteristics of a discipline SLA
1 A discipline has a particular object of research.
SLA investigates the acquisition of a language acquired after the learner’s mother tongue.
2 A discipline has a body of accumulated specialist knowledge referring to a specific object of enquiry, which is not generally shared with another discipline.
SLA has drawn on a range of other disciplinary areas but has now developed specialist knowledge related to such issues as individual differences in learning outcomes, the trajectory of development in an L2, variability in L2 systems, L1 transfer, the role of input and interaction, and the cognitive and social processes involved in L2 learning.
3 A discipline has theories and concepts that can organize the accumulated specialist knowledge effectively.
SLA has spawned a plethora of theories to explain L2 acquisition although, to date, no widely-accepted general theory has emerged.
4 A discipline uses specific terminologies or a specific technical language adjusted to the research object.
SLA borrows technical terms from its source disciplines but also increasingly coins its own special terminology—as can be seen by a brief inspection of the Glossary in this book.
5 A discipline has developed specific research methods according to its specific research requirements.
SLA has borrowed research methods from other disciplines but has adapted these to its own needs and developed new methods.
6 A discipline must have some institutional manifestation in the form of subjects taught at universities or colleges, respective academic departments, and professional associations connected to it.
There are now departments of SLA in universities while courses in SLA figure regularly in postgraduate programmes. SLA has its own conferences.
Table 12.1 SLA as a discipline (based on Krishnan’s 2009 criteria for a discipline)
However, while a case can be made for viewing SLA as a discipline, it is in many respects interdisciplinary in nature: that is, it draws on the knowledge and methodologies of a range of other disciplines. Like other interdisciplinary academic areas, it is subject to continuous change; it is fragmented and heterogeneous; and it interacts with other disciplines in complex ways. For some SLA researchers, this is problematic as it raises questions regarding the boundaries of the discipline, what constitute appropriate methodologies for investigating L2 acquisition, and what the ultimate goal of SLA should be. Klein (1998) considered SLA among ‘the bottom dwellers in the language science’ (p. 53). This is perhaps overly harsh but it is clear that SLA is far from coalescing to provide a uniform account of how an L2 is acquired.
The boundaries of SLA
What constitute the boundaries of SLA is perhaps the area of greatest contention. As we have seen, SLA was originally positioned as a cognitive enterprise (i.e. language is viewed as a mental construct and acquisition as taking place inside the mind of the learner) and the task facing the researcher as that of identifying the mental processes—such as noticing—that account for how learners extract information from the input and how interlanguage systems develop over time. This perspective emphasizes the universal aspects of L2 acquisition as manifested, for example, in the orders and sequences of acquisition of an L2 grammar we considered in Chapter 4. For some researchers, SLA is necessarily a ‘branch of cognitive science’ (Long and Doughty 2003: 4). For other researchers, however, L2 acquisition is primarily a social phenomenon. In Chapter 9, I examined a number of theories belonging to the ‘social turn’ in SLA. These emphasize the local and situated nature of L2 acquisition and point to its heterogeneity and the inevitability of individual differences in the trajectories it follows. As Larsen-Freeman (2007) pointed out, researchers in these two branches of SLA do not ask the same questions and they draw on different kinds of data. The bifurcation that arose in SLA as a result of these competing accounts of L2 acquisition remains unresolved, although in recent years there have been attempts to find a larger lens that incorporates both perspectives in work on complex adaptive systems and—in particular—in Dynamic Systems Theory (see Chapter 8). Nevertheless, arguments over the boundaries of SLA are far from over and SLA continues to reflect parallel epistemologies. For many SLA researchers (myself included), however, this is no bad thing as, arguably, multiple metaphors—’acquisition’ and ‘participation (Sfard 1998), for example—are needed to account for the full complexity of L2 acquisition.
Research methodology in SLA
As the boundaries of SLA have expanded, so too have the research methodologies employed to investigate L2 acquisition. The methods that researchers have employed vary according to how ‘acquisition’ is conceptualized. In Ellis (2006), I distinguished three different ways of defining ‘acquisition’ and discussed the different ways of measuring acquisition associated with each:
Acquisition as emergence
When acquisition is conceptualized in this way, it is necessary to show that learners can demonstrate they use a particular L2 feature that they could not use previously. The primary way of demonstrating this is to take ‘emergence’ as the measure of acquisition. This has been defined as the use of the feature in at least three different contexts in the learner’s spontaneous production. This approach to investigating acquisition has been employed in research based on Pienemann’s (1998) Processability Theory (see Chapter 4 and Chapter 8) and it is also the principal way of investigating acquisition in the social turn (see Chapter 9).
Acquisition as increased control
When acquisition is conceptualized as increased control, the metric applied is target language accuracy—i.e. the extent to which the learner demonstrates a gain in the ability to use a specific linguistic feature in accordance with target language norms. In experimental studies, this is generally achieved by means of tests. Increasingly, however, researchers have used tasks to elicit more spontaneous language use and then examined the accuracy of specific linguistic forms by examining suppliance in obligatory contexts (see Chapter 4).
Acquisition as progress along an acquisitional sequence
All SLA theories acknowledge that there are developmental trajectories as learners’ interlanguages advance. To investigate these, longitudinal studies involving a series of data collection points are required. The research then plots the use of specific linguistic features over time, identifying when and how change takes place. It is the required method for investigating L2 acquisition as a complex adaptive system.
In order to avoid the comparative fallacy, many researchers reject target-like accuracy as an adequate measure of acquisition, although this continues to be the preferred approach for investigating the effects of explicit instruction (see Chapter 10). The usage-based approaches for investigating acquisition which have become so influential in SLA view acquisition as emergence and then plot the development of specific linguistic features over time. An example of this approach is Eskildsen’s (2012) study discussed in Chapter 4.
A major development in SLA has been the use of meta-analysis to provide summative accounts of the results of experimental studies that have investigated various aspects of SLA. Meta-analysis is a statistical technique used to examine the results obtained from a large number of studies that have investigated the same phenomenon in order to arrive at a generalized statement about the effect of a particular treatment—for example, form-focused instruction or input enhancement—on acquisition. Norris and Ortega (2006) published a collec
tion of meta-analyses and several more have since appeared in the journal Language Learning. Previous chapters in this book have drawn on the results of a number of these meta-analyses.
Increasingly, SLA researchers have sought out methods for investigating the cognitive processes involved in acquisition. For example, in Chapter 7 we saw that researchers interested in the role of interaction have used stimulated recall to investigate what learners attend to in the interactions they participated in. Computer simulations of learning, techniques for measuring brain activity, and eye-tracking have also been used to investigate how input is processed (see Chapter 8). The methods employed by cognitive SLA researchers have become increasingly more sophisticated over time as have the statistical tools used to analyse quantitative data.
Social SLA has employed a different set of methods to investigate acquisition in flight. In particular, they have drawn on conversation analysis to investigate in detail how learners participate in conversations and how such participation assists the development of both linguistic competence and interactional competence (see Chapter 1 and Chapter 9). Social SLA has also drawn on the methods employed in ethnography—for example, the use of diaries, interviews, and learner autobiographies—to explore the social worlds that learners inhabit.