‘Lexical approach’

Introduction

‘Lexical approach’ is a term bandied about by many, but, I suspect, understood by few. What does taking a lexical approach to language teaching mean? What are the principles and tenets behind a lexical approach? What problems will teachers have to face if they wish to adopt a lexical approach?

     For the present purposes, I will be using the term lexical approach to mean that lexis plays the dominant role in the ELT classroom, or at least a more dominant role than it has traditionally, which has largely been one of subservience to ‘grammar’ (Sinclair & Renouf 1988). The approach stresses the necessity of using corpora to inform pedagogical materials and the importance of regularly recycling and reviewing the language taught. I should make clear from the start that my understanding of the term lexical approach is not necessarily the same as Michael Lewis’ (e.g. 1996, 1997, 2000), although I imagine that my take on the principles and problems inherent in implementing a lexical approach probably have a considerable amount in common with Lewis’ own views.

     The article begins with a brief outline of what I mean by the term lexis, before briefly outlining two of the tenets which in my view constitute a lexical approach. The same tenets are then problematized at greater length. Finally, while it is argued that there is still much to be done before a lexical approach is accepted by a majority of practitioners and researchers and integrated into mainstream ELT, I close by claiming that the approach can be seen as having many of the same concerns as state-of-the-art applied linguistics.

The concept of lexis

Language teaching has traditionally viewed grammar and vocabulary as a divide, with the former category consisting of structures (the present perfect, reported speech) and the latter usually consisting of single words. The structures were accorded priority, vocabulary being seen as secondary in importance, merely serving to illustrate the meaning and scope of the grammar (Sinclair & Renouf 1988).

     However, a number of studies (e.g. Altenberg 1990; Erman & Warren 2000; Kjellmer 1987; Pawley & Syder 1983) have shown that the Chomskyan notion of a native speaker’s output consisting of an infinite number of “creative” utterances is at best a half-truth: in fact prefabricated items form a significant part of a native speaker’s spoken and written output. Only this can account for what Pawley & Syder (1983: 193) call the puzzle of nativelike selection: a native speaker’s utterances are both “grammatical” and “nativelike”, and while only a “small proportion” of grammatically well-formed sentences are nativelike, that is, “readily acceptable to native informants as ordinary, natural forms of expression”, these are the sentences which native speakers produce. It would seem, then, that speakers need both a prefabricated, automatized element to draw on as well as a creative, generative one—both “idiom” and “open choice” components (Sinclair 1991).

     Once the importance of prefabricated language is acknowledged, the traditional grammar/vocabulary distinction becomes problematic: as the above studies show, native speakers are prone to using much of the same language over and over again rather than starting from scratch each time they speak/write. For the purposes of this article, therefore, when I use the term lexis I have in mind strings of words which go together (i.e. prefabs and collocations) as opposed to the single words language teaching traditionally called ‘vocabulary’: rather than consisting of a repository of content words, lexis is not easily distinguishable from the concept traditionally labelled as ‘grammar’ (e.g. Singleton 1997). This fuzziness suggests that lexis is more powerful than was once thought, and hence deserves a higher priority in syllabuses.

Principles of taking a lexical approach

principle 1: teach real language, not ‘teflese’; use computer corpora but be corpus-based, not corpus-bound

At the centre of a lexical approach is the insistence on teaching ‘real’ English and a rejection of the ersatz language found in the average ELT coursebook; and indeed a number of corpus-based studies (e.g. Holmes 1988; Hyland 1994; Mindt 1996; Williams 1988) confirm that the language coursebooks teach is “not what people really say” (Lewis 1997: 10), it is “TEFLese” (Willis 1990: vii). Hence it can be argued that the only way to avoid distorting the language with this TEFLese English is for the coursebook writer to access the authentic language via corpora, as opposed to relying on their intuition. It is well documented that intuition (even native-speaker intuition) often fails to accurately reflect actual language in use (e.g. Biber, Conrad & Reppen 1994); in contrast, corpora can instantly provide us with the relative frequencies, collocations, and prevalent grammatical patterns of the lexis in question across a range of genres. In addition, light is shed on lexical variation (cf. Fernando 1996; Moon 1998). To illustrate the point, I draw on data from an earlier study (Harwood 2000) comparing the language found in a native-speaker corpus (the British National Corpus) with the language in a selection of coursebooks. In Bell & Gower’s coursebook (1992: 150), for instance, no variation of the phrase You must be joking is included, giving the learner the impression the form is frozen. However, the BNC includes the following variations:

I says [sic] you’re joking                         You’re flipping joking!

You are joking me?                                  You’re joking

You are joking, aren’t you?                    You’re joking, aren’t you?

You gotta be joking!                                You’re joking, of course

You have got to be joking                      You’re not joking?

You have got to be joking me                You’ve got to be fucking joking

You have to be joking                             You’ve got to be joking!

You must be bloody joking!                  You’ve gotta be joking mate

You must be fucking joking!                 You’ve gotta be joking!

You must be joking                                  You’ve just got to be joking

and while it is relatively simple to use native-speaker intuition to point to the fact that You have got to be joking me, for instance, or I says [sic] you’re joking are relatively untypical examples of variation on this phrase, and hence not worth teaching (especially since, if the learner is familiar with more common variations, they would understand and be able to respond to this variant in any case); and while it is similarly straightforward to determine that variations such as You’ve got to be fucking joking cannot be included in international teaching materials because of their potential to offend, we are nevertheless left with a number of typical, ‘polite’ variations which Bell & Gower’s material fails to cover—and which corpus data has brought to the fore (Harwood 2000: 14-16). However, by dismissing some of the variations as inappropriate and hence not being necessarily constrained by corpora, we are being what Summers (1996: 262) calls “corpus-based but not corpus-bound”.

principle 2: recycle and revisit

Nation (1990: 44-45) concludes that coursebooks’ lack of recycling “provide[s] considerable cause for alarm”, before claiming that lexis should be recycled between 10 and 12 times for higher-level learners, and warning that teaching vocabulary without incorporating the necessary recycling is wasted effort. Similarly, Kojic-Sabo & Lightbown (1999) stress that an EFL learner’s need for recycling/reviewing is perhaps more acute than a non-native speaker who is surrounded by the L1 (i.e. an ESL learner): since EFL learners are not continually surrounded by the target language they cannot be said to benefit from any spontaneous reviewing which may result. Regardless of the amount of ‘practice’ material which accompanies the initial presentation, what is needed is repeated exposure over a given period, as opposed to exposing the learner to the lexis once, ‘practising’ it, and never recycling it again (Harwood 2000; Lewis 1997).

Problems of taking a lexical approach

problem 1: corpora and teaching ‘real’ english

Real English, corpora, and ‘learner overload’

The case of you must be joking discussed previously illustrated that corpus data requires adjustment before it can be allowed to serve pedagogical ends: for instance, untypical or culturally inappropriate items will need to be removed from the handout which is given to the learners. However, it is likely that further adjustments will still be required, due to teachability and learnability factors: that is, since anecdotal evidence (and common sense) suggests there is a limit to the number of items learners can learn at any one time (i.e. in a single lesson), including every lexical variant at every opportunity will complicate the issue unnecessarily. Learners will be overwhelmed and will fail to learn any—or at least learn fewer—lexical strings less well than if they had been presented with a smaller, more manageable list in the first place. So implementing a lexical approach requires a delicate balancing act: on the one hand, the teacher will wish to consult the appropriate corpora to avoid the ersatz English of the textbooks which reflects little of the language’s lexical variations and predominant patterns. On the other hand, however, teachers will be anxious to ensure learners are not “overloaded” with too much lexis which would result from exposing the class to as many lexical strings as the corpus describes (cf. Cook 1998). We should be aware of the dangers of teaching learners unusual or deviant variations when, since we have neither the time (as teachers) or the space (as materials writers) to include more than some variations in our lessons or coursebooks, we should ensure these are the variations which will be most useful to the learners. Hence intuition regarding both linguistic and pedagogical matters needs to be exercised: in addition to asking ourselves whether any of the attested corpus examples are untypical in the skill/genre we are attempting to teach (e.g. academic writing: the research paper), pedagogical judgements such as the accessibility/difficulty/volume of material also require reflection, since, while corpora can tell us much, pedagogical concerns such as these are clearly not addressed by the data (Cook 1998).

Some additional limitations of corpora

Although corpora are no pedagogical panacea (e.g. Cook 1998; Widdowson 2000), I do not believe that corpora in themselves necessarily make the implementation of a lexical approach problematic. The key issue is rather how corpus data is selected and manipulated. To take one example of the potential misuse of data, there is a popular but mistaken belief that the frequency with which lexis occurs in a corpus will determine its priority in our syllabus. In fact, I would suggest that the more advanced the learners’ level, the more apparent it becomes that something more than frequency counts is required. Although much has been made of Willis’ (1990) assertion that the most frequent 700 words of English constitute 70% of text, the problem of what one should teach subsequently remains. As Willis’ figures show, this is much less easily prescribable:

The 700 most frequent words cover 70% of text, but coverage begins to drop rapidly thereafter. The next 800 words cover a further 6% of text and the next 1000 words cover 4%…It is true that general frequency is not the sole criterion [for identifying the appropriate lexis for a syllabus]. (Willis 1990: 47)

Hence, while the frequency factor should not be ignored in our attempts to mirror real English in the classroom, it is clear that frequency should not be the only, or even the principal, factor in determining the lexis to teach. Relevant also is work on text type (e.g. Biber et al. 1994) and genre analysis (e.g. Bhatia 1993; Swales 1990), showing that a research article, for instance, will feature different types of structures and phrases when compared with a business letter; and that to a certain extent such features are predictable. So we would do well to bear in mind learners’ wants and needs (cf. Biber et al. 1994): it is evident that the materials designer will have to consult very different corpora when designing materials for pre-sessional postgraduate learners enrolled in English-medium universities who need to develop their academic writing skills, for instance, compared to an intermediate-level general English group who wish to explore some of the most common ways native speakers open a conversation with their peers.

     In summary, corpora in no way constitute a pedagogical “quick fix”: while corpora should undoubtedly stand at the centre of a lexical approach, the teacher and materials designer will need to be aware of the many variables which will influence corpus selection and data manipulation.

From printout to handout

The materials designer needs to acknowledge that there is likely to be a degree of learner (and teacher) resistance to corpus-based materials if the data is handled insensitively, due to the fact such materials are untraditional and also because, more generally, some perceive computers (and therefore computer-based learning) negatively. Such resistance will, of course, only increase should an impenetrable amount of corpus data be simply reproduced straight onto the textbook page (Cook 1998; Leech 1998; Widdowson 2000).

     Hence the requirement for the designer and/or teacher to “do” something with the data. One example of what should be done, if communicative language teaching is to be believed, is to ensure the learner feels involved, investing something of themselves in the material (e.g. cf. Allwright 1981; Coady 1997; Sökmen 1997). The materials designer will need to present the (potentially impersonal) corpus printouts in such a way as to stimulate the learners’ personal involvement (cf. Aston 1995); and while various researchers have been developing Johns’ (1991) practical ideas for exploiting corpus printouts in the classroom for some time (e.g. Fox 1998; Lewis 1997; Milton 1998; Thurston & Candlin 1998; Willis 1998) the same ideas are crucially lacking in published commercial materials (Harwood 2000; Moon 1997). Although this may have been more excusable in the past, when corpus-based descriptions were harder to come by, these days designers’ over-dependence on introspection and intuition is less and less justifiable (Harwood 2000).

Existing published materials are not corpus-based

Since there is evidence that designers are failing to exploit corpus data to shape coursebooks’ lexical syllabuses, the teacher who wishes to push lexis up the agenda on their course is obliged to produce their own corpus-based materials. This constitutes a serious difficulty for the spread of a lexical approach: however willing the individual teacher may be to teach lexically, their institution may well prevent them from doing so (Baigent 1999). In addition, of course, time restraints and an excessive workload result in many teachers introducing only a minimum of their own material onto a course. All of this suggests that the influence of a lexical approach will be negligible while there continues to be a dearth of available published material which abides by its tenets.²

Corpus access

I close this section on corpora and a lexical approach by supposing that, in spite of the difficulties described above, a teacher wishes to consult the appropriate corpora to design lexically based materials. Assuming the teacher has access to the necessary computing technology (a considerable assumption—most teachers around the world do not have such access), they will still be faced with the fact that ELT publishers refuse to grant them access to consult many corpora. Other corpora, such as the British National Corpus, require access fees that the teachers’ institutions may be less than willing to provide. And while it is true that there are now cheap corpora available such as the BNC Sampler, teachers of EAP who require soft and hard science sub-corpora to help students write across the disciplinary spectrum will continue to be denied access because of publishers’ commercial interests.

problem 2: teaching and learning real english

Whatever the problems involved in accessing the appropriate corpus data, a more fundamental concern is whether it is desirable to even choose to teach “real” (i.e. nativelike) English. The question is obviously enormously complex, and I limit myself to sketching out four related issues.

Respecting learners’ wishes

There is evidence to suggest that many learners have no wish to learn real lexis and sound like an L1 (e.g. Anglo-American) user: despite the fact that many teachers (consciously or unconsciously) hold the nativelike model up as the “ideal”, the learners’ non-native variety can constitute a separate cultural identity, marking the L2 speaker out from the native community (cf. Beneke 1981; Carter 1998; Dellar 2000; Hinnenkamp 1980; Kasper & Blum-Kulka 1993; Littlewood 1983; Prodromou 1996; Wray 1999). Meanwhile the Anglocentric coursebook continues to predominate, which presumes a degree of integrative motivation on the part of the learner and implicitly denies or devalues local Englishes (Beneke 1981; Prodromou 1988). (Such problems, of course, necessarily involve questioning the “global” strategy of the major ELT publishers: if local varieties of English, which differ considerably, are to be the target, then perhaps publishers should be more concerned with marketing local, rather than international, “one product fits all” coursebooks. This is discussed further below.)

     In sum, however “real” the lexis is, teachers cannot assume that learners will be prepared to learn lexis simply because “native speakers say it” or “it’s in the coursebook”.

Perceptions of ‘real lexis’

Perhaps some of the objections to teaching real language arise from our perceptions of what exactly real lexis is: I suspect that to many teachers it consists of what Leech (1998) tactfully calls the “less admirable features” of language, which we may not wish our learners to reproduce.³ Alternatively, other teachers may bring to mind the various idioms and idiomatic phrases course and resource books periodically dig up which could be described as parochial and of limited relevance to the class: Hobson’s choice; to send someone to Coventry (Harwood 2000). Yet I would claim that such language, however “real” it may be, is not the kind of lexis which a teacher would be contemplating teaching by following a lexical approach: if learners’ needs remain to the fore, real lexis does not have to be impolite, irrelevant or outlandish. As we saw above when discussing corpus data, to identify a piece of lexis as authentic is not sufficient justification for including it in the syllabus: what is essential, then, is to prioritize (real) lexis according to need.

Non-native teachers

One of the objections to emphasizing real lexis is due to the fact that the majority of English teachers worldwide are non-natives. How well-equipped are non-natives to teach “real” current British slang? In any case, prioritizing this kind of language would reinforce the alleged supremacy of the native speaker teacher as “expert” (Prodromou 1996). However, take what is perhaps a more common example: the non-native speaker teaching an EAP class. In this case, the real lexis in question should present little problem for the non-native teacher who is well-versed in the conventions of academia and is therefore no less expert than a native speaker.

Varieties of real English

Another key question is what we mean by real English and real lexis: do we mean American, Australian, British, Caribbean, Indian, or South African English, to name but a few varieties? Are we striving to perpetuate an educated or less educated model? Where does the English of the millions of speakers of English as a second or foreign language fit in (Hyde 1998; Leech 1998; Prodromou 1996)? Again, I would suggest such questions are best answered by attempting to address the learners’ needs and wants, although it is well for the teacher to bear in mind that issues like world Englishes and intercultural pragmatics are complex: being acutely aware that real lexis will vary immensely depending on the user should help ensure the classroom atmosphere is not one of small-minded prescriptivism.⁴

problem 3: recycling in practice

Coursebooks fail to recycle lexis systematically

A recent study of 12 upper intermediate and advanced coursebooks found that none of the prefabricated language or metadiscourse examined was systematically recycled (Harwood 2000). As Littlejohn (1992) claims, ELT materials are failing to keep pace with applied linguistics research, which in this case would suggest that recycling should be a standard feature of the coursebook. And while it should be conceded that recent studies of lexis (e.g. Sanaoui 1995) have emphasized the importance of the learners managing their own vocabulary learning by means of skills the teacher has helped them develop, I do not believe this exonerates materials (or those who design them) from responsibility for recycling. Rather, I would contend that many teachers in fact underestimate the part recycling plays in language learning, and that the coursebook should engage in recycling to underline its benefits to learners but also to remind teachers to incorporate recycling into their lessons regularly.⁵

Coursebooks have a role to play in encouraging teachers to recycle

However highly teachers rate the importance of recycling, in many classrooms they have little power to ensure it features regularly. Since many institutions worldwide oblige their teachers to follow the coursebook slavishly (cf. Baigent 1999; Dubin & Olshtain 1986), how can the teacher be expected to recycle if recycling activities are left to their whim and are not included in the material?

Recycling needs to consist of more than “doing the same thing twice”

A possible explanation for the apparent reluctance of materials designers (and teachers?) to recycle sufficiently can be found in Lewis’ (1997: 51) assertion that “ “Doing the same thing twice” is still widely considered time-wasting and potentially boring”. As Lewis implies, while a recycling/revisiting strategy should be at the heart of a lexical approach, it is also vital that teachers and material writers ensure recycling is done in an interesting and refreshing way, so that the learners’ interest is still engaged. Variety and novelty, rather than rote learning and staid predictability, should be the cornerstones of the recycling component in a coursebook.

problem 4: face validity for teachers and learners

Learners’ and teachers’ perceptions

I wish to reiterate that it is essential that a lexical approach is implemented with sensitivity by the teacher: it is not a case of throwing out all established pedagogy. With this in mind, I now turn to the question of face validity. Although the term is normally associated with the field of language testing, where it is used to examine how acceptable and credible a test is to users (e.g. Alderson, Clapham & Wall 1995), for this article I take it to mean what learners and teachers expect to devote time to in the language classroom. I will now attempt to illustrate the importance that face validity has when utilizing a lexical approach, and potential difficulties which may arise which can be traced to worries about face validity.

     While material which takes a lexical approach can be built around many ‘conventional’ design principles which feature in more traditional ‘grammar-based’ exercises, where there is material which is not conventional, not the stuff of the standard ELT coursebook, the question of face validity is likely to arise, since teachers and learners will not be used to the materials and may well therefore question their validity. Because all materials feature a “hidden agenda” (Nunan 1989), with what the writer sees as being the essential things to be learned coming to the fore, by its very prominence, lexis is implicitly ascribed an unprecedented degree of importance. But will the class accept this? Might they not demand ‘grammar’ in the sense in which it is normally presented? While we have seen that the Chomskyan generative paradigm cannot be claimed to describe language adequately and that the realms of grammar and lexis are neither readily definable nor even necessarily discrete (and hence in teaching lexis one can simultaneously be teaching grammar), this is not to say that many, or even the majority, of teachers and learners would accept this and be prepared to attach a higher priority to the acquisition of lexis. The prudent course of action, then, is not to abandon grammar teaching in the traditional sense, but to ensure that syllabuses and materials include both lexis and grammar (cf. Wray 2000). We should remember that there is a type of ELT which predominates in many parts of the world which is radically different in its underlying assumptions (i.e. it values traditional grammar instruction more highly) when compared with the state-of-the-art Anglo-American type (cf. Anderson forthcoming).⁶ Whatever the strengths and weaknesses of these more traditional approaches, and however sound the case for a more lexically-oriented approach to teaching may appear, we must proceed carefully if those teachers and learners who see structural grammar teaching as key are to be at all persuaded of the merits of a lexical approach. Cook’s (1998: 60) insistence that a preoccupation with lexis will “inevitably” lead to “a bewildering refusal to teach grammar” on the part of the teacher must be proved mistaken if face validity is to be maintained. Perhaps one of the reasons the COBUILD course was not particularly successful was that teachers and learners had never seen anything like it before, and face validity became an issue. In contrast, the most recent coursebook associated with the lexical approach (Dellar & Hocking 2000) contains a traditional “grammar” component and does not appear unduly different to the standard coursebook. In sum, then, the way to assuage teachers’ and learners’ fears of a lexical approach is to avoid an iconoclastic call to abandon all grammar activities. We should instead simply call for the teaching of lexis to come higher up the agenda.

Implementation

Some of the difficulties I have raised concerning the practical implementation of a lexical approach can be connected to face validity. As Thornbury (1998) has pointed out, teachers are unlikely to be interested in a set of pedagogical principles per se: it is only when the same principles can be applied to classroom situations that their worth is evident. Given the lack of guidance available in the literature at present as to how a lexical approach should be implemented, then, the approach is unlikely to be adopted until it is seen by teachers as operationalizable. So although Lewis (1993) gives us an insight into the kind of syllabuses he does not favour and a range of classroom activities which bring lexis to the fore (Lewis 1997), we are never presented with a comprehensive syllabus based around a lexical approach that Lewis does approve of (Thornbury 1998). Difficulties such as these which hinder the implementation of a lexical approach necessarily involve face validity: it may seem that either (i) the lack of available commercial materials means the approach is misguided, and that lexis is not so important after all; or (ii) that however legitimate teachers and learners believe the approach might be, the lack of materials makes implementation impossible.

     The lack of lexically based materials is now discussed further.

problem 5: the world of elt publishing

The lack of available pedagogical material claiming to take a lexical approach can be used to critique the ELT publishing world. The conservatism of the industry is well documented: ELT publishers fail to respond to findings in applied linguistics research quickly, and indeed often never apply these findings (Littlejohn 1992; Thornbury 1998). In order to maximize profits, materials are developed for the global market, despite the fact that the many varieties of international English being spoken suggests that products should cater for individual local markets instead (Prodromou 1988). All of this helps to explain why at the time of writing, with the exception of Dellar & Hocking (2000), Powell (1996), and the COBUILD series, coursebooks purportedly built around any sort of lexical approach are conspicuous by their absence.

Conclusion: a lexical approach and the state of the art

While I have tried to outline what I see as a number of difficulties regarding the implementation of a lexical approach in this article, I wish to emphasize that I am in no way inimical to the approach per se. Hence I close by pointing out that in many ways a lexical approach shares the concerns of the most current research in a number of areas of applied linguistics.

     Take, for instance, a lexical approach’s insistence on abandoning the misleading grammar/vocabulary dichotomy which has continued to inform ELT materials. The fuzziness of the grammar/lexis distinction is also currently being underlined by studies in phraseology (e.g. Altenberg 1998; Gläser 1998; Howarth 1996); while the emphasis on the importance of prefabs in a lexical approach is confirmed by work on formulaic language (e.g. Aijmer 1996; DeCock 1996, 1998; DeCock et al. 1998; Granger 1998; House 1996; Moon 1997, 1998; Wray 1999, 2000; Wray & Perkins 2000) and metadiscourse (e.g. Crismore, Markkanen & Steffensen 1993; Hyland 1998a,b, 1999; Intaraprawat & Steffensen 1995; Mauranen 1993a,b). If a lexical approach is implemented appropriately, learners will acquire lexis suitable for their needs, a priority which accords with the recognition of the importance of genre analysis (e.g. Bhatia 1993; Swales 1990), in that research in this area shows clearly that the lexis which is suitable for EAP groups, say, may not be so suitable for conversation classes. Hence a lexical approach recognizes that, in order to design material for an EAP class, it is necessary to consult an academic, rather than a general English, corpus.

     As it stands at present, the concept of taking a lexical approach to teaching is work in progress (Thornbury 1998), since there are two main areas connected with the approach which are in need of clarification: while some researchers (e.g. Cook 1998; Thornbury 1998) have critiqued the approach’s purported lack of principled foundation, there is also concern about the practicalities of the approach’s implementation (e.g. Baigent 1999; Lewis 1997; Thornbury 1998). It is hoped that this article has made a contribution to the discussions on both these issues.