The distinction between implicit and explicit knowledge is of great significance for language teaching. The theoretical models emerging from SLA research have taken up differing stances on the 'interface' between implicit and explicit knowledge in the L2 learning process. With reference to these stances, two points of particular interest to L2 classroom instruction are: which type of knowledge contributes more effectively to learning and which type of teaching, explicit or implicit, provides more assistance to the L2 learning process.
This paper discusses some of the influential theories of implicit and explicit knowledge; how the two types of knowledge contribute to learning; and the impact of theory on classroom instructional methodologies. Although both types of knowledge can refer to different aspects of language, this paper focuses particularly on grammar for two reasons: firstly due to space limitations, and secondly because of its importance to language pedagogy.
The paper starts by defining the two types of knowledge and providing an overview of what the corresponding learning and teaching aspects of this knowledge entail. The theories that are then discussed have been grouped according to their stance on how the two types of knowledge interface. Along with a brief description of the theories I will also look at their implications on classroom instruction. The paper concludes by reviewing form focused instruction, which is a good example of how ideas emanating from theory have influenced teaching approaches by integrating the artificiality of learning into a more natural process.
2. Defining the implicit/explict dichotomy
2.1 Implicit knowledge, learning and instruction
Implicit knowledge is commonly associated with a learner's linguistic competence (Ellis, 2005a). Literature on psychology and SLA research uses several overlapping terms to refer to this knowledge, for example, unconscious knowledge, intuitive knowledge/awareness, epilinguistic behaviour, spontaneous/ automated knowledge, or procedural knowledge/rules/memory.
Bialystok (1981) offers the following description of implicit knowledge:
The general form in which information is represented allows us to know things intuitively without being aware of the formal properties of that knowledge. For example, we know a great deal about language that defies mental examination, but the knowledge is demonstrated by our ability to produce correct, coherent utterances.
Implicit knowledge underlies the fluent language skills usually associated with native speakers (Hulstijn, 2007), who have an ability to notice grammatical errors without necessarily being able to explain the rules causing them. Ellis (1994) suggests that this knowledge can be broken down into two sub-categories: formulaic knowledge consisting of pre-fabricated chunks of language; and rule-based which consists of general and abstract structures which have been internalised. Both these sub-categories are stored unconsciously and only become apparent when the language is produced in communication (ibid). Within the brain, implicit knowledge is not restricted to one specific area, but is spread over different regions of the neocortex (Paradis, 1994).
Implicit learning is the forming of implicit knowledge, and is a natural process of acquiring new knowledge unknowingly, and in such way that the knowledge is difficult to verbalise (Ellis, 1994). For example, a learner may unwittingly learn a grammatical rule while working on a meaning focused activity, or notice a structural pattern during a short-term memory task. This learning takes place automatically whenever information is processed receptively, and once the process is initiated, the learner cannot choose not to encode the input (Hulstijn, 2007).
Classroom instruction is considered implicit if rules are not presented and learners are not required to attend to forms (Norris and Ortega, 2000). Examples of implicit instruction include high frequency input, interaction, and recasts (Spada, 2010). Grammatical and lexical resources are a means to an end, and considering the general consensus that development of implicit linguistic knowledge results in language acquisition, the ultimate aim of classroom instruction should be to facilitate this development (Ellis, 2005b). Even though there is still disagreement on how implicit knowledge is acquired, it is generally accepted that communicative activities play an essential role in the process; therefore communicative tasks could be an effective instructional tool when the language learning focus is on implicit knowledge (ibid).
2.2 Explicit Knowledge, learning and instruction
Explicit knowledge refers to different aspects of language, including grammatical, phonological, lexical, pragmatic and socio-cultural (Ellis, 2005a). As with implicit knowledge, several overlapping terms have been used to refer to L2 explicit knowledge, for example, language/metalinguistic awareness, analysed knowledge, conscious knowledge, declarative knowledge, learned knowledge, or metagrammar. Ellis (2004: 244) gives an extended definition of explicit knowledge as:
Explicit L2 knowledge is the declarative and often anomalous knowledge of the phonological, lexical, grammatical, and sociocritical features of an L2 together with the metalanguage for labelling this knowledge. It is held consistently and is learnable and verbalisable. It is typically accessed through controlled processing when L2 learners experience some kind of linguistic difficulty in the use of the L2. Learners vary in the breadth and depth of their L2 explicit knowledge.
Explicit learning is a conscious, deliberate process of structuring explicit, verbalisable knowledge, which can take place while learning concepts/rules in the classroom or it may be initiated independently (Hulstijn, 2007), for example, when a learner refers to a grammar book to find the past participle of a particular verb. This type of learning requires a degree of cognitive development, therefore it is unlikely to take place in early childhood (ibid). Explicit knowledge is said to reside, or at least processed in a specific area of the brain (the medial temporal lobe, including the hippocampus), which is separate to the areas where implicit knowledge is stored (Ullman, 2001).
Norris and Ortega suggest that explicit instruction exists along a continuum, from instruction which is more to that which is less explicit (Norris and Ortega, 2000). A deductive approach to classroom instruction is an example of a 'more' explicit from of instruction, where rules are explained before a structure is presented in context. An example of a less explicit form is inductive instruction, here learners are asked to attend to and make metalinguistic generalisations on a form which has already been presented in context (ibid).
3. The implicit/explicit interface
The contradictory claims regarding the dichotomy of implicit and explicit knowledge have focused mainly on how these two types of knowledge interface. The interface hypothesis presents three positions which argue the extent to which explicit knowledge is involved in L2 acquisition.
3.1 The non-interface position
At a certain age children stop using their language to communicate and begin to look at it reflectively (Tunman and Herriman, 1984). Children's use of explicit knowledge is considered to exhibit different levels of consciousness depending on their literacy skills, whereas their acquisition or use of implicit knowledge shows little variation (Ellis, 2004). This implies therefore, that knowledge which initiates conscious or explicit linguistic behaviour is distinct from that unconscious or implicit knowledge which characterises natural language use.
Krashen (1982), Paradis (1994) and Schwartz (1993) claim that "acquired" and explicitly "learned" L2 knowledge does not interface, arguing that the former is responsible for language fluency, and the latter is only useful to monitor communicative output. Paradis (1994) also rejects the possibility of explicit knowledge converting directly into implicit knowledge, or vice versa. He suggests that since these two types of knowledge exist in neuroanatomically distinct memory systems, they can interact but transfer of knowledge from one to the other is unlikely (ibid).
Krashen (1982) argues that formal teaching of grammar is unnecessary as it has no effect on language acquisition, and explicit L2 knowledge may never actually convert to implicit knowledge. In addition he suggests learners have little ability to learn grammar. Similarly, Truscott (1996) adds that the only benefits of formal grammar instruction are in preparing learners for assessment which look to measure explicit metalinguistic knowledge rather than communicative ability.
In his input hypothesis Krashen states that like L1, L2 acquisition is also a natural process which occurs implicitly while a learner is exposed to comprehensible L2 input (Krashen, 1982). If learners are motivated, they will naturally follow an in-built syllabus to develop their inter-language, and using an intuitive process of trial and error, eventually acquire the L2 (Ibid). Krashen's theory was the inspiration behind the natural and communicative approaches to language teaching.
Counter to Krashen's claims, evidence from research has shown that despite immersion in the L2, learners continue to make grammatical errors. A study by Harley and Swain (1984), for example, showed that immersion students relying purely on comprehensible input were unable to achieve high levels of language proficiency.
Effect of L1 transfer is a possible reason why implicit learning processes are less effective for L2 (Ellis, 2008). Unlike a newborn infant, the L2 learner's neocortex is already configured and optimised for the L1 (ibid). L2 processing and automatisation therefore occur non-optimally, as they have to rely on implicit L1 representations (ibid).
A weaker form of the non-interface position suggests a possibility of implicit knowledge being transferred to explicit knowledge through conscious reflection and analysis of implicitly generated output (for example, Bialystok, 1982). Similarly, Ellis (1994) also argues for a seperateness of the two types of knowledge, he proposes a connectionist account of implicit knowledge as a complex interconnected network which is neurologically detached from explicit language knowledge. However, he suggests that the two types of knowledge may be derived from each other and that they can interact during language use (ibid: 235).
3.2 The interface position
The non-interface position has been attacked both theoretically and empirically by other SLA researchers, who have addressed the role played by explicit knowledge in language acquisition. Sharwood Smith (2004), for example, uses the interface hypothesis to argue that explicit knowledge can be gained from implicit knowledge, and similarly explicit knowledge can be transformed to implicit knowledge using: contextualised communicative practice, repeated use and corrective feedback.
Taking a strong interface position, the skill building theory (DeKeyser, 2003) suggests that a procedularised form of explicit knowledge is functionally equivalent to implicit knowledge when learners are given plenty of opportunities to engage in meaningful communicative practice. This practice is an essential step in proceduralising the target language for spontaneous use; hence it is important that learners are motivated to engage in this process through non-threatening feedback (Faerch, 1986).
Many studies have provided empirical evidence to justify the role of explicit grammar teaching. Ellis (1994), for example, has shown that explicit language instruction leads to faster learning, and that adhering to an implicit focus on meaning fails to provide high levels of competence. However, for grammar instruction to be effective, some researchers have found that a careful selection and sequencing of rules is essential, as well as a determination of the learner's linguistic readiness to accept a new grammatical item (Ellis, 1994; Fotos, 1994).
The grammar translation and cognitive approaches, which were popular in the 1960's and 70's are typical examples of explicit teaching methods. These methods were influenced by the belief that an explicit knowledge of grammatical rules precedes their use (Ellis, 2008).
The PPP model is another instructional approach taking an interface stance. PPP emphasis a focus on form, and stipulates that a language feature should be: explicitly presented, then practiced and finally produced in order to procedularise the feature. Swan (2005) sees PPP as a useful approach for presenting and practicing language structures under semi-controlled conditions. However, PPP is now widely seen as lacking a firm basis in SLA theory, its linearity and behaviourist nature fails to take into consideration the stages of developmental readiness that a learner goes through (Ellis, 2003); and its systematic instructional approach is unlikely to lead to acquisition of the language feature taught (Skehan, 1996).
3.3 The weak interface position
A weak interface position proposed by some theorists, suggests the possibility of transferring knowledge between the implicit/explicit systems. Two popular processing models from cognitive psychology which take a weak-interface position are McLaughlin's (1987) information processing model, and Anderson's (1983) ACT model.
The information-processing model proposes that complex behaviour evolves from simple modular processes that can be isolated and analysed independently (McLaughlin, 1987). Within this framework L2 learners use controlled processing, requiring a lot of 'attentional control' to generate language sequences, which are then stored in short-term memory (ibid). Through repeated activation, these sequences become automatic and are transferred to long-term memory, where they can be accessed with minimal 'attentional control' (ibid).
Based on a similar viewpoint, the ACT model (Anderson, 1983), suggests that declarative knowledge (knowing that something is the case) leads to procedural knowledge (knowing how to do something). Three types of memory are defined in this framework: a working memory (similar to short-term memory), and two types of long-term memory, declarative and procedural. Anderson maintains that during learning, declarative knowledge becomes procedural and automatised, and that both types of knowledge are stored differently (ibid). A learner might start of by studying a rule (for example, "Use a and an when the following word starts with a consonant or vowel, respectively"), but every time a phrase containing this rule is produced or received, the phrase is stored as an instance in memory (Logan, 1988). Increasing encounters with these instances raises their activation levels to such an extent that eventually retrieving a stored instance will be quicker than applying the rule (ibid).
Other versions of the 'weak interface' position also support a possibility of knowledge transfer but set restrictions on when and how this transfer can take place. Pienemann (1989), for example, argues that learners cannot transmit knowledge between the two systems until they are ready to acquire the linguistic form.
Ellis (1994) claims that explicitly teaching declarative rules can have a 'top-down' impact on perception, leading to saliency of the taught language features. Learners are then able to notice the feature during input, and by comparing it with their output can consciously 'notice the gap' (ibid). Explicit knowledge in this case acts as a stimulus in activating conscious awareness and the subsequent storage in long-term memory (Ellis, 2005b). The significance of explicit knowledge in this case is not so much as a contributor to acquisition, but as a detector of specific language features in the input.
Ellis suggests consciousness raising (CR) as a way of setting a linguistic focus to tasks, and encouraging learner autonomy by requiring learners to derive explicit grammar rules independently (Ellis 2005b). CR raising tasks can be inductive or deductive, in the former learners are expected to induce an explicit representation of a rule, whereas in the latter the rule is provided at the beginning of the task (Ellis et al., 2003). The main aims of CR tasks are to involve learners in goal-orientated communication and to encourage the development of explicit knowledge (ibid).
4. Form focused instruction (FFI)
FFI consists of a number of approaches to teaching that advocate a focus on both meaning and form (for example, Doughty and Williams, 1998; Lightbown and Spada, 1990). The distinction between the various types of FFI is that some are implicit in nature, and others are more explicit. Ellis (2001) defines FFI as a type of instruction which includes "any planned or incidental instructional activity that is intended to induce language learners to pay attention to linguistic form". Long (1991) distinguishes between two types of FFI: focus on formS (FonFs) and focus on form (FonF). The former involves teaching discrete grammar points according to a synthetic syllabus leading to a preselected linguistic target (ibid). FonFs is regarded as an explicit form of FFI (Housen and Pierrard, 2005) and is more in line with the interface position. A typical example of a FonFs approach is the PPP model.
FonF on the other hand is a more implicit form of FFI (ibid), and aims to "overtly draw the students' attention to linguistic elements as they arise incidentally in lessons whose overriding focus is on meaning or communication" (Long, 1991: 45). FonF is based on the idea that first and second language acquisition are similar in that they both rely on exposure to comprehensible input from natural interaction (ibid). However it also takes into account important differences: that learners cannot acquire many of the grammatical aspects of language through exposure alone, and that this needs to be balanced by providing a focus on grammatical as well as communicative aspects of the second language (ibid). The instructional activities associated with the FonF approach involve a mixture of implicit and explicit techniques, for example: input enhancement where a target form is highlighted for awareness; and a structure-based task (Fotos, 2005) which requires the completion of a meaningful task using the target form, before the latter is explicitly taught and practiced further.
Long (1991) contends that FonF instruction may be more effective than a focus on meaning (FonM) or a FonFs approach, because it is more consistent with the findings of SLA research. From a psycholinguistic perspective a FonF teaching in the classroom is justified for three main reasons:
FonM may be useful in developing oral fluency, however it fails to provide high levels of linguistic or sociolinguistic competence (Ellis et al., 2003). The FonM approach is based on Krashen's (1981) hypothesis which states that all that is needed to acquire a language is extensive exposure to rich comprehensible L2 input. However, while researching literature comparing instructed with uninstructed learning (FonM), Long (1991) found that instructed learning was much more effective in achieving high proficiency levels.
A FonFs approach is based on the idea that classroom L2 learning is derived from cognitive processes and therefore involves the learning of a skill (Ellis et al., 2003). However empirical evidence (for example Pienemann, 1989) suggests that a FonFs approach does not guarantee that learners will develop the ability to restructure their interlanguage. Studies have shown that classroom learning follows a sequentially similar acquisition process as natural learning (Ellis et al., 2003). However, in the classroom learners may follow an inbuilt syllabus, allowing them to benefit or acquire only those aspects of FFI for which they are linguistically ready (ibid). Therefore deciding on which language feature learners are ready to acquire may pose a difficulty for FonFs instruction.
A FonF instruction draws attention to the target form through a contextually meaningful communication, allowing the learners to develop their fluency and accuracy (Ellis et al., 2003). FonF is pedagogically efficient in that it can focus specifically on those language features that either need clarification or are problematic at a contextually relevant moment (ibid).
FonF also gives an opportunity to learners to receive feedback in a meaningful context, allowing them to 'notice the gap' between their interlanguage and the negative evidence provided by the feedback (ibid). Johnson (1996), in his skills building theory suggests that feedback is most useful for learners when it is presented in 'real operating conditions'. Corrective feedback exposes learners to the correct form and encourages them to produce it themselves; leading to a possible acquisition of these forms (Ellis et al., 2003).
The above discussion supports the efficacy of FonF instruction; however the effectiveness of this method in some EFL contexts is doubted. In educational contexts where teachers are obliged to follow a tightly controlled syllabus, or where class size does not permit individual feedback, a FonF instructional approach may be difficult to implement (Poole, 2005). What this suggests is that pedagogical implications of SLA studies on implicit and explicit knowledge need to be related to different learning and teaching contexts.
This paper discussed some of the prominent theories that have emerged from SLA research on implicit and explicit knowledge. As well as defining the two types of knowledge, their impact on the learning process and instructional practices were also highlighted. A look at 'focus on form' instruction showed how the ideas from different theoretical viewpoints have merged to give a teaching approach which balances both implicit and explicit learning.
Although research has shown that traditional explicit grammar instruction is unlikely to lead to the implicit knowledge needed for proficiency in a language, there is still a lot of controversy regarding the best alternative (Ellis, 2006). The conflicting views on the overall role of implicit and explicit knowledge in SLA point to the complexity of the issue and suggest that a thorough understanding is still evolving. Ellis (2008) suggests that because consciousness and linguistic knowledge are so difficult to conceptualise and operationalise, improving our insight in these areas is a major challenge. In order to help gain a deeper understanding, future research needs to collaborate with developments in other disciplines such as cognitive science and cognitive neuroscience (Doughty and Long, 2003).