3.1 Things taking over

Exciting developments inspired by new research interests can come to fill the language-teaching horizon. This happened with audio-lingual teaching, when views on learning as habit-formation, valuable in their place, sometimes gained a stranglehold on language teaching overall. In the early days of the communicative approach, some classes did little besides work on functions and communicative activities. Research on learning and communication strategies in the 1990s sparked off a virtual pedagogic cottage industry, with Oxford (Reference Oxford1990: 24–25), for example, recommending that teachers and students internalize and work from a complex and daunting categorization of strategies which would have left little time for anything else. The ‘lexical approach’, based on a precarious analogy with the emergence of first-language grammar, attracted more attention than it probably deserved in the 1990s, and its focus on formulaic sequences often led to an overemphasis on teaching ‘chunks’ (Swan Reference Swan2006). Corpus research has encouraged some teachers to bring corpus material into the classroom, with the risk of requiring their students inappropriately to become amateur language analysts. Pure linguists have not always respected Hatch's (Reference Hatch1978) principle that the only question a researcher should answer is the one that he or she has asked, and some have not been slow to step outside their specialities brandishing pedagogic recommendations. The eminent corpus linguist John Sinclair, for instance, told teachers, quite unrealistically, to present learners only with corpus-attested examples (Reference Sinclair and Wichmann1997: 30ff). Functional grammatical models such as Cognitive Grammar or Systemic Functional Grammar are often recommended as appropriate for pedagogic application, despite the unsuitability of any such systematic models for teaching purposes (Swan Reference Swan and Simpson2011: 565).

3.2 Misapplied linguistics: a disconnect

It is not enough for a teaching approach to conform to current SLA thinking, however persuasive the supporting arguments. It must also fit the relevant context, offering solutions that match the problems inherent in the situation. I feel, however, that there is sometimes a disconnect in this respect.

Task-based language teaching (TBLT) is often promoted as being preferable in general to ‘traditional’ approaches involving proactive language syllabuses (Willis Reference Willis1996: 133–137; Skehan Reference Skehan1998: 93–94; Ellis Reference Ellis2003: 319–320; Ellis & Shintani Reference Ellis and Shintani2014: Ch. 6; Long Reference Long2015b passim). However, there is a considerable mismatch between the problems central to many instructional contexts, and the solutions offered by teaching that is exclusively or primarily based on a syllabus of tasks. TBLT tends to focus on consolidating what has already been partially acquired – what we used to call, over-simply, ‘fluency practice’ – through communicative tasks in which learners develop their command of English while ‘[making] use of their own linguistic resources’ (Ellis Reference Ellis2003: 16). New linguistic material relevant to the tasks emerges or is (up to a point) supplied as needed, and may be acquired during task performance. However, many of the world's learners have little out-of-class exposure to the language they are studying and little time in class – 100 hours a year is not an uncommon pattern. In such situations, it is mathematically impossible for unstructured ‘naturalistic’ input and primarily meaning-based classroom communication to supply and adequately consolidate all the grammatical and lexical elements that are most needed (Swan Reference Swan2012: 63). These can only be presented and internalized through careful selection, prioritizing and (where necessary) focused practice, and consequently by means of appropriate formal and semantic syllabuses. Communicative tasks will naturally support the consolidation and activation of syllabus-based input, as they always have done, and task-related research has enriched our repertoire of such activities and our understanding of how to exploit them. But hard-core task-based teaching (unlike task-supported teaching) does not fit well with the main requirements of learners in this kind of situation, however appropriate it may be to certain other contexts. (For detailed discussion, see Swan Reference Swan2005.)

3.3 Straw men

Input-poor instructional situations are not then, in my view, well served by teaching approaches based on the theory that acquisition takes place primarily as a by-product of communication. Approaches that are in principle better adapted to such contexts have, however, frequently been condemned in the SLA literature on the grounds that they are ‘discredited’ and have ‘failed’ (e.g. Skehan Reference Skehan1998: 94; Ellis Reference Ellis2003: 207). ‘Discredited’, in this connection, seems to mean ‘incompatible with my views’; we are not generally told by whom or for whom these approaches have been discredited, or where, when or how seriously they have failed. Straw-man caricatures of ‘traditional’ instruction (often meaning any teaching other than TBLT) are frequent. Robinson (Reference Robinson and Robinson2001: 291) portrays such teaching as relying on ‘simplified classroom language and texts which are functionally and linguistically impoverished, prohibiting exposure to forms and functions learners may be ready to learn, or need to use’. Willis & Willis (Reference Willis, Willis, Carter and Nunan2001: 173) and Long (Reference Long2015a: 567) offer similar thumbnail sketches. Both Ellis (Reference Ellis2015: 244) and Long (Reference Long2015b: 349) use ‘PPP’ (the presentation–practice–production sequence often used in grammar teaching) as shorthand for ‘traditional approaches’. This is rather like equating the whole of driving instruction with, say, lessons in overtaking. While no doubt teaching of the kind pilloried does exist, these descriptions bear no relation whatever to well-designed syllabus-informed instruction, as a glance at the major published courses of the last 30-odd years will show. The impression is frequently given of a paradigm shift, in which unsatisfactory older methods with no theoretical foundation have been replaced by totally different approaches based on robust theory and research. This is quite misleading.

There is of course nothing new about this kind of thing. Grammar-translation and audio-lingual teaching, whatever their real defects, have been characterized for decades in over-simple dismissive terms. The contrastive analysts such as Lado and Fries (from whose work language teaching benefited greatly) were seriously misrepresented by their successors such as Dulay, Burt & Krashen (Reference Dulay, Burt and Krashen1982), who criticized them both for saying things that they did not say, and for not saying things that they did. The misrepresentation is routinely repeated in present-day references to their work; see Swan (Reference Swan2007a) for an attempt to set the record straight.

3.4 The invisible practitioner

In general, the theory-practice gap is narrowing. Lightbown, in her ‘Anniversary Article’ (Reference Lightbown2000), refers to the growth in classroom-based SLA research over the previous 15 years, and this growth has continued unabated, to our distinct benefit. Nonetheless, the invisible practitioner is still a familiar character – or non-character – in academic writing, and I am disturbed when researchers fail to acknowledge the work carried out on my side of the theory-practice divide. This can happen, for instance, in discussions of language course and syllabus design. Most of the world's millions of language learners follow syllabuses that are instantiated in textbooks. These are largely created by experienced professionals whose work calls on specific knowledge and skills, building on a developing tradition established by their predecessors over long periods. And yet it often seems as if these thousands of creative practitioners, perhaps because they do not often contribute to learned journals, are effectively invisible in the academic world – although names such as Alexander, O'Neill, Abbs, Freebairn or Soars have probably become known to more people worldwide, with good reason, than those of almost any contemporary novelist. An article on the role of materials in the language classroom (Crawford Reference Crawford, Richards and Renandya2002), for instance, makes no mention at all of the textbook as a selection and presentation vehicle. Several of the chapters on cognition and second language instruction in Robinson (Reference Robinson and Robinson2001) touch on language syllabus design. Of the 1,000-plus references at the end of the book, six are to work by language-teaching practitioners; none of these is a course designer. The authors of a recent book reporting on their progress in drawing up grammatical specifications for the different levels of the Common European Framework of Reference for English (Hawkins & Filipović Reference Hawkins and Filipović2012) make no reference at all to the vast body of relevant work carried out by pedagogic grammarians over half a century. It may be that, coming from another field (grammatical theory), they are simply unaware that such work exists. Alternatively, they may just have felt that the principles informing their research make it necessarily more valid than the approaches of their humble practitioner colleagues. This is not the case (Swan Reference Swan2014): their provisional findings amount to a small sample of what has long since been well charted by practitioners, and their methodology is of a kind that, to borrow a phrase from Meara (Reference Meara1993), ‘should not be recommended to numerate readers with high blood pressure’.

Sometimes the practitioner is not so much ignored as treated with what amounts to contempt. Ellis, in one of his various defences of TBLT (Ellis & Shintani Reference Ellis and Shintani2014: Ch. 6), divides its critics into two groups. One group, researchers, ‘[bases] its criticisms on evidence’, and the researchers are responded to with the normal professional courtesies. The other group comprises what Ellis derisively calls ‘pundits’: ill-informed practitioners who, in his perception, ‘rely on opinion derived from their own experience of teaching’, and who use his findings and the associated pedagogic recommendations as ‘aunt sallies’, easy targets for their unsophisticated critical missiles (Ellis & Shintani Reference Ellis and Shintani2014: 152–153). Two things need to be said here. First of all, one does not need to be a researcher to take issue with a researcher's logic, any more than one needs to be an animal to discuss animal rights. The key question is whether the argumentation is valid, not whether the person who questions it is a scholar, a teacher, a violinist or a garage mechanic. And secondly, this is not, in my view, an appropriate way in which to engage in academic debate.

3.5 Hypotheosis

It is not unusual in the social sciences for empirically ill-supported hypotheses or beliefs to harden into doctrine. In our own discipline, the direct method, the audio-lingual approach, the fringe methods of the 1970s and 1980s, and various brands of communicative methodology have all been promoted to the accompaniment of confident assertions that ‘now we know’ how languages are taught and learnt. Such assertions are sometimes little more than expressions of opinion, supported by citations from other scholars who share the same belief, like the strategy of the person in Wittgenstein's celebrated analogy (Reference Wittgenstein1953: §265) who buys several copies of the same newspaper and uses each to confirm the accuracy of the others. Where evidence is offered, the argument may involve questionable extrapolation from a limited empirical base, robust though the research findings may be in their own right.

In recent years, much of SLA discussion has revolved around a group of hypotheses to do with ‘noticing’ (Schmidt Reference Schmidt1990), ‘interaction’ (Long Reference Long and Winitz1981, Reference Long, Ritchie and Bahtia1996), incidental ‘focus on form’ (Long Reference Long, Lambert and Shohamy2000), ‘output’ (Swain Reference Swain and Lantolf2000) and ‘processability/teachability’ (Pienemann Reference Pienemann1998). Stated in their weak forms, these ideas are relatively uncontroversial. Few people would contest that:

• • noticing formal features may have some importance for acquisition

• • interaction may generate feedback, which may trigger acquisition

• • pushed output may lead to the consolidation of known material

• • acquisition may occur via incidental focus on form during communication

• • some second language morphemes and complex structures may be acquired (and therefore teachable) in predictable orders or stages.

The hypotheses have generated considerable research investigating how far, under what circumstances and for what language elements they apply, and this has added greatly to our knowledge.

However, I have found myself concerned when strong and over-generalized forms of these hypotheses have been enunciated and presented as justifying firm and wide-ranging pronouncements about instructed acquisition, with the implication – and even the direct assertion – that the pronouncements are validly grounded in research or theory. The reasonable observation that ‘things seem to happen like this under certain circumstances’ can transmute into a declaration that ‘this is the only (or principal or most effective) way in which things happen’. SLA may then be said to take place solely or primarily through communication as mediated by interaction, noticing, incidental focus on form and so on; confident references may be made to what is said to have been established by research, or to the cognitive processes through which, we are told, acquisition necessarily takes place (e.g. Willis Reference Willis1996: 13; Ellis Reference Ellis2003: 210). This cognitive-interactionist model of instructed SLA has figured largely in the theoretical support for TBLT (e.g. Long Reference Long2015b). However, empirical evidence for the strong forms of these hypotheses and the claims built on them necessarily has limited scope, and replication studies are rare. It is not clear, therefore, that the large generalizations put forward follow logically from the data cited in their support; nor has TBLT been shown generally to result in more successful learning.

To take one of these hypotheses as an example: the existence of natural developmental sequences in the acquisition of morphemes has been appealed to as justifying the rejection of traditional grammar syllabuses as fundamental components of a language course (Doughty Reference Doughty and Robinson2001: 227–228; Ellis Reference Ellis, Hinkel and Fotos2002: 19–20; Ellis & Shintani Reference Ellis and Shintani2014: 73). If there is a predetermined natural order constraining the acquisition of grammar, the argument goes, we cannot set up a teaching sequence for most grammatical items, since we do not know where they fit into the natural order, and cannot therefore determine when learners are developmentally ready for one or the other (Ellis & Shintani Reference Ellis and Shintani2014: 73). Now research by Pienemann and others has indeed confirmed some earlier beliefs about natural acquisition order, showing plausibly that structures which make greater demands than others on learners’ developing processing capacity tend to be acquired later (Pienemann Reference Pienemann1998). And it has been established that the acquisition of some complex structures typically goes through non-canonical stages leading up to native-like use (English negation is the standard example). But this is as far as the evidence goes. We have no reason at all to believe that the learning of most grammatical items is constrained in this way: that for yet-to-be uncovered developmental reasons, students might need to learn comparatives before relative pronouns, dativizing verbs before quantifiers or infinitives of purpose before possessive ’s. To claim that learnability findings preclude the operation of a grammatical syllabus is a large and unjustified leap across a wide logical gap.

The other hypotheses referred to also fail, in my view, to constitute a solid platform for a cognitive-interactionist model opposing the use of proactive syllabuses (Swan Reference Swan2005). Like instructors in other domains, language-teaching practitioners have always used inventories of high-priority learning targets and known difficulties to inform their work, and there appears to be no sound theoretical or research basis for the recurrent doctrinaire assertion that they are wrong to do so.

Large-scale pedagogic recommendations based on unwarranted extrapolation of this kind are effectively ideological in nature. They belong to a category which Richards (Reference Richards, Richards and Renandya2002: 21–23) characterizes as based on ‘theory-philosophy conceptions’: views of what stands to reason, is self-evident, or is manifestly desirable in psychological or sociological terms. The cognitive-interactionist construct, valuable in itself, can thus come to illustrate what Pullum (Reference Pullum1996) calls ‘an epistemological affliction: difficulty in telling the difference between (i) data that are well accounted for if construct C is posited and (ii) evidence that C actually exists’. In any scientific study, it is surely essential to remain aware of the logical gulf between the big things we believe and the small things we can prove.

4.1 The simplicity bias

Perhaps one reason why we oscillate between conceptualizations of language teaching is for the sake of simplicity. Language and language learning are complex, with many different aspects, and the whole business can be difficult to conceptualize and deal with. Pedagogic messages which say ‘It's all . . . . . . . . .’ (habit-formation, lexis, communication or whatever – fill in the blank) have a seductive appeal. However, it is not all any one thing, and I cling to my conviction that no single approach or one-dimensional syllabus – of structures, functions, tasks or anything else – can serve as an effective organizing principle for instructed language teaching. Complexity is what we have to live with.

4.2 The success bias

Because existing methods are not very successful, we can easily feel that they are not working, throw them out and discredit their proponents. But we don't necessarily fail; we just don't usually do very well. Foreign languages are too hard for most people to learn well, especially in input-poor contexts, in the instructional time available. The sense of failure no doubt relates partly to the apparently intractable carry-over problem: the difficulty of achieving fluent and accurate spontaneous production of what is taught, and especially of grammar. Consolidation of the teaching and practice of structures by continuing attention to accurate use during more communicative work is difficult to contrive, and will usually have disappointing results. But switching to the opposite of what we were doing before is not necessarily a solution, even assuming that the problem in question – how to achieve a high level of grammatical accuracy in a second language – is one that can be, or needs to be, solved for most learners.

4.3 The English bias

A good deal of SLA theory grew up from studying the acquisition of English, and a certain bias continues. Now English happens to be a morphologically light language. Much of the basic structure can be picked up easily from exposure and practice, and beginning students can move quickly towards a modest working command. It is interesting to speculate what current language-teaching theory would look like if the main reference language were, say, Russian or Greek. These languages have a formidable array of inflectional grammar, and a beginning or elementary student has to pay a great deal of attention to these forms if he or she wants a moderate level of grammatical accuracy. They cannot easily be picked up from simple exposure to the limited input available to most students, supplemented by brief episodes of focus on form, and there is a powerful case for learning tables of inflections by heart to serve as ‘declarative crutches’ (DeKeyser Reference DeKeyser, Doughty and Williams1998: 49) on the road to procedural fluency. It is not at all certain that the act of faith whereby forms are taken to be learnable largely through classroom interaction would be possible if an inflecting language were the focus of language teaching theory to the extent that English is.

4.4 The ‘language in use’ bias

Language is a formal code, used to convey meaningful messages, and as such it faces in two directions. Consequently, conceptualizations of language and language teaching tend to swing to and fro between the poles of code and message, form and meaning, knowledge and skill, imitation and expression, control and freedom, learning and using. It seems to me that much of applied linguistic thinking, along with theory-informed language teaching, is currently well towards the meaning-expression-freedom-communication end of the pendulum swing, and that this has been the case for several decades.

This pervasive bias has, I believe, skewed our thinking about language itself. Explorations of language in use, of the way different speech acts are expressed, and of the conventions governing verbal interaction and text construction have added greatly to our knowledge. They have also, however, encouraged a common confusion about the scope of pragmatics, leading to a widely repeated belief that language items in general inherently have two kinds of meaning, semantic and pragmatic, and that both kinds need to be taught. Hymes’ declaration (Reference Hymes, Pride and Holmes1971: 278) that there are rules of use without which the rules of grammar would be useless has resounded down the decades. Doughty & Williams suggest ‘that the degree of effectiveness . . . of focus on form ultimately depends on the level of integration of the learner's attention to all three aspects of form, meaning and function’ (Reference Doughty and Williams1998: 244–245). Similar recommendations are made by Celce-Murcia & Larsen-Freeman (Reference Celce-Murcia and Larsen-Freeman1999: 4) and Ellis (Reference Ellis2005). But the belief is, quite simply, a myth. To be sure, all utterances derive their communicative force from the interaction of their grammatical/lexical meanings with the context of utterance. ‘It's Wednesday’ can mean ‘Put the bins out’, ‘It's your turn to babysit’, ‘Remember Granny's coming to supper’, or any number of other things. But this is message meaning, not code meaning; there is no extra pragmatic information that we can give about declarative clause structure or the language items it, is, and Wednesday that will contribute to a better understanding of the utterance in question in context. Indeed, many structural elements in languages lack anything that can even reasonably be called semantic meaning, like the position of English relative clauses, gender agreement in Italian or the verb-second rule in German. For an optimistic attempt to strangle this confusion at birth, see Swan (Reference Swan1985); for a more extended treatment of the issue, see Swan (Reference Swan2007b).

In this context it is particularly irritating for a pedagogic grammarian to be told, as we often are, that the proper teaching of meaning, ‘language in use’ or ‘grammar as choice’, is a recent discovery.

Note the interesting implication that in the bad old days we did not even teach semantic meaning most of the time, concealing from our students the dangerous information that plural nouns refer to more than one entity, that past tenses are prototypically used to refer to past time, or that forms like older and more beautiful express comparison.

Perhaps the ‘two kinds of meaning’ myth is not really a cause for concern: few teachers are likely in practice to spend time trying to distinguish for their students the pragmatic from the semantic meanings of, say, microwave, waterproof or third-person -s. But the impact on course design and methodology of the ‘language in use’ bias is a different matter, and I believe it has seriously skewed our instructional principles and practice. ‘Communicative’ teaching has largely filled our horizons for 40-odd years, to the detriment of approaches and procedures which do not fit the paradigm. In our professional discussions, activity-related concepts are virtually guaranteed automatic approval. ‘Learner-centred’, ‘meaning-based’, ‘holistic’, ‘discourse’, ‘discovery’, ‘process’, ‘interaction’, ‘negotiation’ and ‘strategy’ are good things to say, on a par with ‘motherhood’ and ‘democracy’. ‘Teacher-dominated’, ‘form-based’, ‘discrete’, ‘sentence-level’, ‘transmission model’, ‘product’, ‘syllabus’ (unless preceded by ‘task’), ‘memorization’, ‘repetition’ and ‘drill’ are not so good: they relate to ‘bad’ pedagogic attitudes felt to be discredited and undesirable. At classroom level, our preference for learner-centred, naturalistic, activity-based learning can all too easily lead us to replace teaching language by doing things with it. The programme for the 2016 IATEFL conference listed around 650 presentations. Of these 25 or so were concerned with grammar teaching, 15 with vocabulary and 20 with pronunciation; these were vastly outnumbered by presentations related to communicative activities. In this conceptual climate, teachers can easily suppose that any kind of engagement with ‘real’ language – analysing a text, practising listening comprehension, launching a discussion or working through a communicative task – is automatically productive, without the need for critical scrutiny of the supposed language-learning payoff. For an unsympathetic examination of so-called ‘skills’ lessons from this point of view, see Swan & Walter (Reference Swan and Walter2017).

In the bad old days we were often better at teaching language than at teaching people to use it. We provided students, so to speak, with the necessary bricks and mortar and left them to build their own houses. Today I feel we often fall into the opposite trap: we get learners building houses right away, but assume that the various supplies needed will magically materialize – as if delivered by elves – at the right times and in the right quantities. This is a precarious assumption.

4.5 The need for common sense

Common sense is an unreliable faculty with no academic standing, and I was once roundly told off by a journal editor for attempting to invoke it. However, I think it can reasonably be appealed to in cases where theory seems to be losing contact with terra firma. One of my favourite odd ideas from an earlier generation is Candlin's ‘retrospective syllabus’ (Reference Candlin and Brumfit1984), which could only be constructed after the course had finished. More recently, I have found it hard to take seriously the dismissal of a concern with linguistic ‘products’ as typical of outmoded traditional approaches (Ellis Reference Ellis2003: 29). If I take foreign language lessons, products are exactly what I want, in the form of a reasonable command of the main aspects of the language, efficiently mediated. And I would not wish to come at this primarily by negotiating form and meaning with other beginners in the ways explored by Swain (Reference Swain2002). It has actually become possible, bizarrely, to criticize language teachers for focusing on language, as when Robinson (Reference Robinson and Robinson2001: 292) criticizes ‘language-based approaches’ because, unlike task-based approaches, they often involve simplified language and systematic practice. But what on earth is wrong with that? Would one condemn a piano teacher for giving ‘music-based’ lessons and setting work on scales and simple studies? One also finds teachers being told not to behave like teachers: Willis (Reference Willis2003: 167) takes exception to materials and activities which imply that students should learn from the teacher and follow his or her directives. Teaching can indeed go too far in the direction of form focus and teacher control; but if language teachers are not supposed either to teach or to concern themselves with language, perhaps we need to backtrack a little.

Some branches of 21st-century theory seem, indeed, to be in danger of losing touch with language teaching altogether. Allwright (2003: 114) tells us that according to the principles of ‘exploratory practice’ we should ‘above our concern for instructional efficiency, prioritize the quality of life in the language classroom’. Ellis & Shintani, in their characterization of TBLT at the level of method, say that ‘no attempt is made to specify the linguistic content to be learned’ (Reference Ellis and Shintani2014: 33). I would not wish to entrust my own quality of life to a heart surgeon whose training had paid no attention to the planned coverage of appropriate professional knowledge, no matter how happy a time he or she had had in medical school. Kumaravadivelu (Reference Kumaravadivelu2006: 201), speaking from some distance around the social turn, enunciates ten ‘macrostrategies’ which characterize ‘post-method’ language teaching. They include ‘maximize learning activities’, ‘activate intuitive heuristics’, ‘promote learner autonomy’, ‘ensure social relevance’ and ‘raise cultural consciousness’. These and his others are fine slogans, but I find it difficult to see what they have to do with language teaching. In the face of pronouncements like these (for which I would suggest that ‘punditry’ is a not inappropriate word), perhaps we need to remind ourselves that the main business of language teaching is, actually, teaching language.

In the belief that satire can be a more effective counter to this kind of thing than reasoned argument, I have published occasional papers poking fun at some of the more imperfectly earthed currents of thought that circulate in our field. Many of these appeared in various issues of the BAAL Newsletter; some are reprinted in Swan (Reference Swan2012). There is always a danger, unfortunately, that a satirical intention may pass the reader by. An article in the ELT Journal on teaching English by sensory deprivation (Swan & Walter Reference Swan and Walter1982), written to lampoon the fringe methods that mushroomed in the 1970s, elicited more than one serious request for further information about the approach, and for some time the title showed up occasionally in other writers’ reference lists.

4.6 The future

Applied linguistic research and theorizing will of course continue to provide essential support for the work of language-teaching practitioners. I feel, however, that these undertakings could sometimes benefit from a commitment to standards of argumentation and proof closer to those that are common in physical enquiry, to the extent that this is possible in the social sciences. Greater receptivity on the part of journals to replication studies would help in this respect (this journal is a notable exception to the general pattern). If this can be achieved, and if the problems and biases discussed above can be overcome, I believe it should be possible to make rapid progress. Given our knowledge of both formal and functional aspects of language, our growing understanding of acquisitional processes, and our impressive array of methods and materials, we probably already have most of the necessary ingredients for a balanced and effective model of instructed second-language learning. The pedagogic landscape is changing very quickly with the IT revolution, but it seems to me that the underlying principles governing language teaching are unaltered. Our job is still essentially to:

• • find out what knowledge and skills our learners need

• • subtract what they already know via their mother tongue or from earlier learning

• • subtract what they can get outside the classroom

• • of what is left, establish which elements matter most

• • of those, establish how many can be effectively taught and learnt under the instructional conditions in the time available

• • teach these by the methods that are most appropriate for each.

Defining the most appropriate methods is the difficult part, but that is what applied linguistics is for. It has my full confidence.