2. The role of instruction in SLA

The field of SLA was initially influenced by research in first language (L1) acquisition research and mentalist theories of language acquisition. There were three major areas of inquiry: accuracy orders of grammatical morphemes, acquisition of grammatical features and the identification/explanation of L2 errors. Many of the early SLA studies were carried out with children, adolescents and adults who learned their L2 in the informal setting – that is, in the schoolyard or in the workplace in interaction with other native (and non-native) speakers. The overall findings from this research showed that L2 errors are similar to those made by children learning an L1, that there are similar sequences of development for morpheme accuracy and other grammatical features and that the primary source of errors was L2 development, not L1 interference (for overviews of this research, see Gass & Selinker Reference Genesee and Beebe2008; Spada & Lightbown Reference Spada, Lightbown and Schmidt2010; Lightbown & Spada 2013). The conclusion drawn from this research was that although the processes of learning an L1 and an L2 were not identical, they were very similar, thus adding support to mentalist views of language learning. However, this research was carried out with naturalistic learners, not instructed learners, and this is an important distinction in the SLA literature.

Naturalistic SLA occurs in informal settings (such as the playground or the workplace) where learners are expected to ‘pick up’ the L2 through observation and interaction with others in much the same way that children acquire their L1. Instructed SLA occurs in formal settings (such as classrooms) where language is intentionally taught and intentionally learned – usually within a limited amount of time. Such an experience is quite different from that of L1 acquisition. Research on instructed SLA began in earnest in the early 1980s, looking at the effects of instruction on accuracy orders and developmental sequences and the effects of instruction on the rate of development and ultimate attainment. The results from the available studies of instructed SLA at that time led Diane Larsen-Freeman and Michael Long to conclude in their 1991 book An introduction to second language acquisition research that formal instruction does not alter acquisition sequences in any significant way, but that it does affect the rate of development and has positive effects on ultimate attainment (Larsen-Freeman & Long Reference Lightbown1991). However, these conclusions were based only on a handful of studies at the time and they were mainly descriptive, not experimental, in nature.

Nonetheless, some researchers concluded on the basis of the similarities observed between L1 and L2 developmental sequences that it was not necessary (or desirable) for teachers to focus learners’ attention on language forms through instruction or corrective feedback, because L2 learning, like L1 learning, could be fostered through exposure to meaningful input and an emphasis on the exchange of messages rather than on language forms. Even though several SLA researchers pointed out that it was premature to apply these findings to L2 pedagogy, there was no shortage of applications (or misapplications) to L2 instruction at that time. This included the ‘Natural Approach’ (Krashen & Terrell Reference Lapkin, Harley and Lapkin1983) advocating comprehensible input as sufficient for L2 development, and the Procedural Syllabus (Prabhu Reference Schwartz1987) echoing many of the assumptions of Krashen's acquisition/learning distinction, both of which fall under the umbrella of the strong version of communicative language teaching, which claims that instruction and corrective feedback are not necessary for L2 learning (Johnson Reference Kramsch1982; Spada Reference Spada2007). Despite dissenting voices, this movement spread like wildfire in the 1980s, had a tremendous impact on L2 pedagogy, and has continued up to the present time in some contexts. Theoretical arguments that corrective feedback is not helpful and may even be harmful also continue to this day and are represented, for example, in the work of John Truscott, who argues that corrective feedback contributes to ‘learning’, not to ‘acquisition’ (Truscott Reference Truscott1996, Reference Truscott1999), a point that is related to the third issue that I will address later in my talk – the nature of implicit and explicit L2 knowledge in SLA.

It is important to make clear that it was not just SLA research that led to this misapplication to L2 pedagogy. At the same time, there was considerable fatigue with the audio-lingual method and a strong desire on the part of teachers and curriculum designers to move from structure-based approaches to more communicative approaches to L2 instruction. To be sure, the introduction of communicative competence by Dell Hymes (Reference Ioup, Boustagui, El Tigi and Moselle1972) also impacted L2 pedagogy, but there is little doubt that SLA research added legitimacy to the pendulum swing that took place in L2 teaching in the late 1970s and early 1980s.

Finally, after a couple of decades of instructed SLA research, evidence accumulated against the notion that comprehensible input is sufficient for L2 learning. The findings indicated that instruction and corrective feedback are beneficial and necessary for L2 learning in classroom settings (Spada Reference Spada, Cummins and Davison1997; Norris & Ortega Reference Ortega, Hinger, Newby and Unterrainer2000; Ellis Reference Ellis2001). Of course, there is still much that we have to learn about the relationship between instruction and L2 learning, including – to name but a few – whether different types of L2 instruction and corrective feedback (i.e. more or less explicit/implicit) are more effective than others, whether certain aspects of language are more amenable to instruction, and whether different characteristics of learners interact with type of instruction to produce different outcomes. Nonetheless, there is compelling evidence that a combination of form- and meaning-based instruction leads to greater L2 success than an exclusive focus on either one. That is, neither an exclusive focus on form typical of structure-based approaches to teaching (e.g. grammar translation and audio-lingual methods), nor an exclusive focus on meaning as reflected in the ‘natural approach’, is advised. Furthermore, there is increasing evidence that instruction and corrective feedback can contribute not only to learners’ analyzed grammatical knowledge but also to their ability to use language spontaneously in communicative interaction (Spada Reference Spada, Lightbown and Schmidt2011; Lyster, Saito & Sato 2013).

The overextension of findings from SLA research based on ‘naturalistic’ L2 learning to instructed L2 learning resulted in a ‘perfect storm’ of SLA theory and the desire for pedagogical change. If these findings based on naturalistic SLA had not been extended to instructed L2 learning, the pendulum swing from overwhelmingly form-based to exclusively meaning-based approaches to L2 instruction might have been avoided or at least diminished. But of course we will never know. Nonetheless, it would have been prudent if the words of Evelyn Hatch – to apply SLA findings to L2 pedagogy with caution – had been heeded at that time. Now I'd like to move to another example of where the findings of SLA research have been misapplied to L2 pedagogy – this time focusing on the role of age in L2 learning.

3. The role of age in SLA

One of the things that most people agree about when it comes to L2 learning is that early starters are likely to attain high levels of linguistic competence, often comparable to native speakers of the language, whereas late starters (adolescents or adults) are less likely to reach comparable levels. Late starters exhibit much greater variability in their levels of linguistic attainment and the majority of them would not be mistaken for native speakers of the language. Of course there are exceptions to this, and several cases have been reported in the literature (Ioup et al. Reference Johnson and Newport1994; Birdsong Reference Birdsong2006).

One explanation that has been offered for the differences between early and late starters is biological: that there is a critical period after which it is no longer possible to learn language in the same way or to achieve the same level of competence as someone who learns before that period (Scovel Reference Shiu and Spada1988; Long Reference Lyster, Saito and Sato1990; Singleton Reference Spada2001); this has become known as the Critical Period Hypothesis (CPH). In the literature, different views have been expressed as to the duration of this period, but, generally speaking, it ranges from two to fifteen years of age, depending on particular aspects of language (e.g. pronunciation, grammar). And the argument is that if language is not learned some time during this ‘window(s) of opportunity’, it will not be complete (Lenneberg Reference Larsen-Freeman and Long1967).

In a review of empirical studies investigating the role of age in first and second language acquisition in the 1970s, three important generalizations were made that helped to distinguish between the rate of learning and ultimate attainment: (1) adults proceed through early stages of syntactic and morphological development faster than children (where time and exposure are held constant); (2) older children acquire faster than younger children – again, in the early stages of syntactic and morphological development where time and exposure are held constant; and (3) acquirers who begin natural exposure to the L2 during childhood generally achieve higher L2 proficiency than adults (Krashen, Long & Scarcella Reference Krashen, Long and Scarcella1979).

However, subsequent research comparing the native-likeness of younger and older starters has been mixed, and agreement is far from complete. For a long time the most robust evidence for the existence of maturational constraints in SLA appeared to come from a study by Johnson & Newport (Reference Johnson and Newport1989), who investigated the grammaticality judgments of Korean and Chinese learners of English and found that those who began learning English the earliest achieved the highest scores. However, these findings have been questioned on methodological and empirical grounds and new research has cast doubt on neurobiological explanations for younger learners’ advantages (e.g. Bialystok & Hakuta Reference Bialystok, Hakuta and Birdsong1999; Birdsong & Molis Reference Birdsong and Molis2001). Furthermore, as indicated above, there is evidence for native-likeness among some post-puberty starters (Ioup et al. Reference Johnson and Newport1994).

Even though the CPH itself does not have unanimous support at present, there does seem to be agreement that early starters have a long-term advantage for ultimate attainment and that late starters have a short-term advantage for learning rate. However, this consensus is based on findings in which the language learner is completely surrounded by the target language, that is, L2 learners who are immersed in the language in the natural environment. What about instructed learners in second/foreign language classrooms? What does the research say about this group of learners in relation to the question of age?

As early as 1975, Clare Burstall compared British students who started to learn French as a foreign language at different ages in school and found that there were consistent advantages for older learners (Burstall Reference Burstall1975). In the 1980s, research investigating the age factor in Canadian French immersion programs revealed that when late-immersion learners (i.e. secondary school) were compared with early-immersion learners (i.e. primary school) there were advantages for late starters on some aspects of their learning (e.g. reading), and they did just as well as early starters on other measures (e.g. listening comprehension) (Genesee Reference Hatch1988). In a large-scale study of core French programs in which learners received instruction for approximately 40 minutes a day, the results on all language measures showed no differences between learners who started later and those who started earlier (Lapkin, Hart & Harley Reference Lapkin, Harley and Lapkin1998).

The most recent and arguably most extensive research to examine the age factor in relation to instructed L2 learning is that of Carmen Muñoz and her colleagues, who are investigating the learning of English as a foreign language (EFL) in schools in Spain. What is particularly valuable about this program of research is that the researchers have carried out longitudinal studies following the same children over several years in primary and secondary schools. They have observed (among other things) that the older EFL learners progress faster than younger learners on a variety of oral and written tasks including oral fluency, vocabulary acquisition, morphological accuracy and writing (Muñoz Reference Muñoz2006, Reference Muñoz2011). They have also observed that younger learners catch up with older learners on oral tasks but not on written measures. Furthermore, the younger learners do not surpass the older learners over time, as has been observed in the research on age with natural L2 learners. That is, the older learners started out stronger and even though the younger learners improved along the way, the older learners maintained their advantage throughout.

Robert DeKeyser (Reference DeKeyser2000) proposed a revised version of the CPH to explain these findings, suggesting that the hypothesis applies only to implicit learning that is activated in a naturalistic setting (with massive exposure), not to explicit learning that is activated in school settings. This suggests that the use of implicit and explicit learning mechanisms is context-dependent. Thus, younger learners in school contexts may be largely deprived of their potential advantage when there is not enough exposure and contact with the language for L2 learning to proceed in the same way as L1 learning. Furthermore, school contexts favor older learners, because of their superior cognitive development and the advantages provided by their explicit learning mechanisms (Muñoz Reference Norris and Ortega2011).

Again we see the importance of context in determining the relevance of SLA research findings for L2 pedagogy. Unfortunately, the advantages that have been found for older learners over younger learners in the instructed SLA context do not seem to have had much influence when it comes to making decisions about when to introduce L2 and FL instruction in the schools. For example, Ministries of Education throughout the world are today revising their teaching policies/practices by moving the teaching of EFL from the secondary to the primary level. There is little doubt that this is because the English language has become a valuable commodity for countries wanting to position themselves more firmly in the global economy. The decision to introduce English language instruction at earlier grade levels is based on the commonly held assumption that when it comes to learning an L2, earlier is best! However, once again, these decisions are based on research that has investigated age and L2 learning in the natural, not the instructional, setting. And the assumption that ‘early is best’ is deeply ingrained and difficult to budge.

I had a recent experience that was a poignant reminder of these deeply held beliefs when I was invited with some colleagues to travel to a non-English speaking country to speak with teachers, teacher educators and representatives from the Ministry of Education about whether they should start EFL instruction in grade 4 instead of grade 6. We spent two full days giving papers and workshops, and participating in round table discussions with these stakeholders about research on the role of age in L2 learning. We described research showing that learners who start learning a language early in the school years do not succeed more than learners who start later when exposure to the target language is restricted to the school setting. We also described several studies showing that when the amount of instruction is limited (i.e. to two or three hours a week), there are no linguistic advantages for early starters over late starters.Footnote ¹ We suggested other ways of organizing and distributing the instructional input to children that have been shown to be effective in other school contexts: for example, intensifying the instruction over shorter periods of time later in the school year (Collins et al. Reference Collins, Halter, Lightbown and Spada1999; Collins & White Reference Collins and White2011). We were well received and our time with this group of educators was positive, but the latest information we have is that a decision has been made to begin EFL in grade 4!

Let me quickly say that there are other reasons why decisions like these are taken, and even though input from SLA researchers was welcome, it was clearly trumped by other factors – factors that are more political and economic in nature. Nonetheless, here is another example of the misapplication of SLA research: taking findings from studies on ‘naturalistic’ SLA and misapplying them to the L2/FL classroom.

The two examples that I have just talked about I consider to be misapplications of SLA to L2 pedagogy because, in both cases, the changes were concrete, visible and tangible (i.e. methods of L2 instruction were modified and implemented differently; major changes were made to the organization of the school curricula to offer language instruction earlier). The last area of SLA research that I would like to discuss is not easily characterized as a misapplication of SLA research but is probably best discussed in terms of its relevance (or dare I say irrelevance) to L2 pedagogy.

4. Implicit and explicit knowledge in SLA

As I indicated at the outset of my presentation, not all of SLA research needs to be applicable to L2 pedagogy: some areas of inquiry address basic questions about SLA. The distinction between implicit and explicit L2 knowledge is one of them. However, because this distinction has received quite a bit of attention in the instructed SLA literature and implications are drawn for L2 instruction based on these distinctions, I would like to offer some thoughts and raise some questions about the relevance of this distinction for L2 pedagogy.

I am motivated to talk about explicit and implicit L2 knowledge based on my own research investigating the effects of different types of instruction on these knowledge types and the frustrated reaction that often results when this topic comes up for discussion in my graduate SLA classes. In the many years I have taught this course the discussion and debates around this distinction are always lively, and in many cases strong resistance has been expressed about the usefulness of such a distinction. I will describe this shortly, but first let's take a look at these constructs and how they have been defined and investigated in SLA research.

Explicit knowledge is described as analyzed, conscious and declarative: for example, knowing that a sentence is ungrammatical and, in some cases, to be able to explain it with reference to a grammar rule. Implicit L2 knowledge is considered as unanalyzed and intuitive: for example, being able to judge the same sentence as ungrammatical but unable to explain why. Thus, explicit knowledge is often described as knowledge with awareness; implicit knowledge as knowledge without awareness (Ellis Reference Ellis2005; Hulstijn Reference Hulstijn, Doughty and Long2003, Reference Hymes, Pride and Holmes2005). In the SLA literature, particularly literature influenced by a Universal Grammar-based linguistics perspective, implicit knowledge is considered to be primary because it is thought to represent a learners’ true competence – the underlying internal grammar, similar to Krashen's (Reference Krashen1981) notion of acquired knowledge.

As most of you know, the conceptual framework that has influenced discussions of how instruction is related to implicit and explicit knowledge is the interface/non-interface contrast. According to the non-interface position, implicit knowledge only develops through exposure to large amounts of naturally occurring input and opportunities for meaningful interaction in much the same way that children acquire their L1, whereas explicit knowledge develops from intentional learning of language rules or patterns. Thus, the claim is that explicit (i.e. not implicit) knowledge results from instruction (Schwartz Reference Scovel1993) and, importantly, the knowledge systems remain separate without the possibility of one type of knowledge converting to the other. There are two versions of the interface position, each taking a different point of view. According to the weak interface position, explicit knowledge cannot become implicit knowledge. However, learners’ explicit knowledge may help them to notice differences between their output and the instructional input, leading to changes in their implicit knowledge. With practice, learners can achieve the ability to retrieve and use this knowledge so quickly and easily that, in performance, they appear to be operating from implicit knowledge. The strong interface position holds that explicit and implicit knowledge are not distinct entities, and that conversion of explicit knowledge to implicit knowledge takes place through practice. The argument is that instructed L2 learners start out with explicit declarative knowledge of a particular language feature (that is, taught via rule presentation) and that, through subsequent practice in meaningful communication, this knowledge becomes proceduralized and eventually automatized for the L2 learner (DeKeyser Reference DeKeyser and DeKeyser2007).

It is important to keep in mind the theoretical distinctions between implicit and explicit L2 knowledge when considering the results from the instructed SLA literature that point to the benefits of instruction. It has been argued that these benefits may be due to the fact that in most of the instructed SLA studies, learners’ progress has been measured with tests that tap into their explicit knowledge. This became evident in the Norris & Ortega (Reference Norris and Ortega2000) meta-analysis and in a subsequent review of that analysis by Doughty (Reference Doughty, Doughty and Long2003). Since then, there has been an increase in the development and use of measures of implicit knowledge in instructed SLA research, with a growing number of studies showing that instruction has benefits for both types of L2 knowledge (Mackey & Goo Reference Muñoz2007; Spada & Tomita Reference Truscott2010).

One task that is considered to be a promising measure of L2 implicit knowledge is the Elicited Imitation (EI) task. According to Rosemary Erlam (Reference Gass and Selinker2006), this is because it is reconstructive in nature, its focus is on meaning, and it is performed under time pressure. In my own research I have used an elicited imitation task in two studies: (1) an instructional study in which we examined the effects of form-focused instruction on the development of implicit and explicit L2 knowledge (Spada et al. Reference Spada and Tomita2011) and (2) a validation study of an EI task (Shiu & Spada Reference Singleton2012), in which we used factor analyses to compare learners’ performance on seven language measures that were classified as explicit or implicit measures of L2 knowledge according to criteria developed by Ellis (Reference Ellis2005).

In the instructional study we examined the effects of form-focused instruction in relation to learners’ performance on an error correction task (hypothesized as a measure of explicit knowledge) and on a picture-cued oral production task and elicited imitation task (hypothesized as measures of implicit knowledge). The same measures were included in the EI validation study with the addition of two grammaticality judgment tasks (one oral and one written), also hypothesized as measures of implicit knowledge, and an expanded version of the error correction task that included correction and explanation sections as measures of explicit (and metalinguistic) knowledge. In the validation study we also administered a retrospective questionnaire to learners asking them what they were thinking about (i.e. form or meaning) when they completed the EI task.

In the instructional study we observed that, on the EI task, the instructed group showed a more significant improvement over time than the comparison group. We initially interpreted this as evidence that the instruction had a positive effect on the development of learners’ implicit L2 knowledge. However, the results of the retrospective learner questionnaire administered in the validation study have led us to question this interpretation. In response to an open-ended item on the questionnaire: ‘What were you thinking about while doing the task?’, 21.1% of the learners reported that they were thinking about the form of the sentences. An additional 19.3 % reported that they were thinking about both form and meaning. This means that 40% of the students were thinking exclusively about form or equally about form and meaning while completing a task that was intended to focus their attention on meaning by putting time pressure on them to access their knowledge rapidly (without awareness) and thus trigger implicit knowledge. However, this does not seem to be what happened. It looks as if the learners were able to complete the EI task successfully by accessing their explicit knowledge, or perhaps some combination of implicit and explicit knowledge. Of course these results could be an artifact of the EI task, and more testing and validation of the task is needed. Notwithstanding this possibility, these findings suggest that time pressure and a focus on meaning do not guarantee that learners will access their implicit knowledge. Indeed, it has been argued that no pure measures of explicit or explicit L2 knowledge exist and that ‘researchers will have to content themselves with eliciting knowledge under conditions that are more or less conducive to the retrieval of implicit and explicit knowledge’ (DeKeyser Reference DeKeyser, Doughty and Long2003: 320). This makes intuitive sense, particularly given that some cognitive psychologists see the implicit/explicit distinction as a continuum rather than a dichotomy.

The fact that many of these learners reported that they were able to use the target form quickly and accurately on this measure means that, while it might have looked as if they were accessing their implicit knowledge, it appears that they were, in fact, rapidly accessing their explicit knowledge. This ability is likely to be more evident with learners who, like the learners in our study, have experienced explicit grammar-based instruction over many years, and through practice have proceduralized their explicit knowledge to the extent that it has become automatic and thus indistinguishable from implicit knowledge – at least on the surface. This raises questions about how relevant the distinction between explicit and implicit knowledge is for L2 pedagogy, given that most instructed learning, particularly with older learners, is conscious, declarative and deliberate.

As I indicated earlier, when explicit and implicit L2 knowledge come up for discussion in my SLA graduate courses it always results in lively discussion and debate. All of my graduate students are L2 or FL teachers and the majority are bilingual or multilingual, so English is their second, third or fourth language. They are all highly proficient in English, many of them are near-native speakers and writers of English and most of them have learned English in an instructional setting. When they read claims that implicit linguistic knowledge is the primary goal in SLA, that explicit and implicit knowledge are separate systems and that explicit knowledge does not convert to implicit knowledge, they express frustration with the distinction between the two knowledge types and the hypothesized relationships between them. Based on their own experiences as highly successful instructed L2 learners and teachers of L2s/FLs, they do not understand why implicit knowledge is given such importance in SLA. This leads us to discussions of Chomsky's (Reference Chomsky1968) notion of linguistic competence, his distinction between competence and performance and to Krashen's acquisition/learning hypothesis and the arguments that acquisition (i.e. implicit knowledge) is primary and that explicit teaching/learning does not lead to acquisition.

While my students acknowledge the theoretical relevance of these distinctions and the need for empirical work to explore the extent to which both types of L2 knowledge are distinct, how they develop, and where they are represented in the mind, as L2 instructors (and learners) they do not find these constructs relevant or helpful when making decisions about how to teach their learners and, in particular, how to assess their progress. What is more meaningful to them is the idea that with practice, learners are able to proceduralize their explicit knowledge, leading to greater automaticity over time. Because this idea is developmental in nature, it seems to resonate more with L2 teachers (and learners) because in the end it doesn't really matter whether their learners’ greater automaticity is representative of implicit knowledge or rapidly accessed explicit knowledge. Furthermore, given that the development of implicit knowledge requires extensive input over long periods of time – two requirements that most L2 teaching contexts do not meet – a more realistic and attainable goal for L2 instructors is to create conditions that will help learners to proceduralize their explicit knowledge. This is even more important in the FL context where few or no opportunities exist for extended input in the target language beyond the classroom setting.