As a preview of our next teacher-training session, a training clinic with Geoff Jordan on the 21st of February, it was about time we had a look back to our last workshop – and indeed our first publicly held SLB training event, “Teaching Lexically”. In fact, Geoff’s “forthright” views on this topic were mentioned during the presentation … could we say that “forthright” is a collocation of “Geoff Jordan”?

Anyway, to kick off on a high, we got some great feedback from the session. 8 out of 10 respondents said the session was “really useful” (the other two said “quite useful”), and some were moved to comment:

“I’d like to congratulate you and thank you for offering this workshop. It made me reflect on my own process as a language learner and how we often forget to use our own experience as a foundation to guide us in our teaching. When training on a TEFL course, we are kind of provided with recipes on how to teach the language skills but there is so much to explore beyond this initial training.”

“My only previous exposure to teaching lexically was two pages and 10 minutes worth at an ESL seminar. I was especially intrigued by the flip-flopping use of L1. Use of L1 at ESL settings in the States is almost considered heresy.”

“Thanks for the workshop. Great location, great chance to network and a great topic I knew nothing about. I really liked the activity at the end to put it all into context.”

That activity, an application of George Woolard’s “messaging” approach to some clips from Seinfeld (for which I owe a lot to Paul Walsh’s Soup Nazi lesson), is available to download and try out here. Read on for some more context to this lesson idea.

A brief history of teaching lexically

We described the development of computerised language corpora as a founding moment in the history of teaching lexically. Staying with this idea, we paused for the first of a series of off-slide activities. Computerised corpora, in that they allow vast amounts of authentic written and spoken language to be examined,  facilitate the easy identification (amongst other things) of a word’s collocations. Could attendees predict what one such corpus, the BNC, would give as the most frequent nouns in the slot “the development of ______”? Try it yourself – answers at the end of the post!

We then discussed what the consequences of this were, in so doing highlighting the phrases or chunks that made up this provisional conclusion:

Prezi capture 1

We qualified that “paradigm shift” was probably a bit strong, as apart from the figures who have taken this notion on further – we reviewed the contributions of Jane and David Willis, Michael Lewis, Michael Hoey, Hugh Dellar then later George Woolard – there isn’t much evidence of a lexical approach in the majority of published ELT materials these days.

However, the participants felt there was something in it, from the point of view of experience at least. I presented a Spanish phrase I had first learnt in Guatemala as an example:

¡Que le vaya bien!

I was a real beginner when I encountered this, and needless to say, were anyone to have explained to me the various grammatical bits that make up this utterance – the subjunctive mood, the positioning of object pronouns and the use of the polite third person form – it would have done me no good whatsoever. But once I had figured out that it meant something like “Have a nice day (sir/ma’am)”, and the correct response was “igualmente”, not only could I understand it, I could actually use and respond to it too. This is in itself a kind of phrasebook learning – which we’ll see again later with Woolard.

From Lewis to Dellar

Having looked at Lewis’s famous pronouncement that language is “not lexicalised grammar, but grammaticalised lexis”, and studying definitions of the terms “nativelike selection“, collocation and colligation via Pawley, Syder and Hoey (references at the end of the post), we turned to a more detailed examination of a more recent application of Lewis, namely Hugh Dellar’s concept of “teaching lexically”, which gave us the title of the session. I decided to stick my neck out and define this term as best I could:

Captura de pantalla 2015-02-08 a las 15.06.36 Captura de pantalla 2015-02-08 a las 15.07.09

We then tried another off-slide exercise – as a way of underlining the point. What else should not be done, or should one not do, “in isolation”? Again, answers at the end of the post! But this led to some consideration of what constitutes practice of lexical chunks, and how retention of these can be facilitated so that “nativelike selection” can be achieved. Perhaps what the lexical approach shares with the rule-based approach is its love of gapgfill exercises …

Captura de pantalla 2015-02-08 a las 15.20.33

But we inserted a note of caution about how such chunks are presented. Participants were familiar with the carpetbombing type of lexical approach sometimes favoured by FCE preparation books, where, for example, 6 phrasal verbs with “come”, some with multiple meanings, are presented and practised on a single page. Do students tend to remember any of these? No, they don’t. We brought in the idea of “interference theory”, where the similarity between items studied tends to hinder retention. Our remedy was to teach such verbs as they come up in material or are required by student conversation, avoid presenting similar-looking batches and pay close attention to the typical collocates of the verbs. If students can successfully notice in which contexts and with which objects or other structures a phrasal verb might repeatedly occur, the greater perhaps the chance of retention. So we asked attendees: What or who is it we usually “come across”, for example? Answers at the end again. There was amusement that “old photos”, the object of “come across” apparently favoured by FCE course writers, did not feature in the BNC top twelve!

Against the lexical approach

At this point we reached something of a crossroads in proceedings. Up until now, we’d been presenting concepts associated with teaching lexically fairly uncritically – it was time to unleash Geoff Jordan! Geoff’s views on this are much better expressed by himself (see the references below). But he isn’t the only one to have had doubts about the lexical approach. We summarised “the problem” thus:

Captura de pantalla 2015-02-08 a las 15.42.40

In other words, frequency of use shouldn’t be the only yardstick for selecting language to teach; Hoey’s theory of “lexical priming”, used by Dellar to prop up his notion of teaching lexically, leaves much to be desired; and since David Willis 25 years ago, no one has offered either a coherent lexical syllabus or a systematic way of teaching one. According to Jordan:

Captura de pantalla 2015-02-08 a las 15.48.37

Jordan, however, does not throw the lexical phrase out with the theoretical bathwater, and leads us to the work of Nattinger and DeCarrico – which may offer some kind of rapprochement between the radical privileging of the lexical chunk on the one hand, and the entrenched and  traditional privileging of the grammatical structure on the other. The diagram below is supposed to represent the continuum between lexis and grammar as described by Nattinger and DeCarrico. The commentary starts from the bottom left (the slide was animated), so Jordan’s comment at the top is supposed to be read last, if that makes any sense:

Captura de pantalla 2015-02-08 a las 15.53.58

So, like most things which become polarised, we usually find the answer somewhere in the middle – or in this case, sliding between the two extremes, recognising the need for a focus on chunks AND on more general patterns.

Which brought us neatly to perhaps the most recent manifestation of the lexical approach, George Woolard’s concept of “messaging”.

The message is the medium

Captura de pantalla 2015-02-08 a las 16.05.24

We looked at how this idiosyncratic methodology mixes the lexical approach with behaviourist-type drills, grammar analysis and translation work, concluding that perhaps Woolard has found an effective way of sliding along the pole between lexis and grammar – pedagogical pole-dancing, if you will – in the presentation, study, practice, retention and production of new language. Woolard admits, however, that messaging may be regarded as “dull”, and may even be nicknamed “The Scottish Approach”! Working against this negative casting of my and George’s nationality, we looked at an application of messaging which incorporates engagement activities and a clip from Seinfeld in an attempt to lift the mood (and lower Krashen’s famous affective barrier). You’re welcome to try it – find the materials here, and the clip is below. The lesson covers the first 4 mins 17 secs.

The lesson activity, which we tried out in part during the session (having some fun with substitution drills), allowed us to present a slightly revised version of the messaging process:

Captura de pantalla 2015-02-08 a las 16.05.51

The participants seemed intrigued both by Woolard’s approach and by his idea of “textsbooks”, coursebooks organised around listening and reading material for which transcripts would be available in multiple languages online. We briefly discussed reasons why translation has largely been excluded from orthodox teaching methodology in recent years – the UK-centric notion that teachers will always be working with mutlilingual groups, for example – before bringing the session to a close.


Our practical conclusions were as follows:

  • Put more focus on lexical chunks, less on grammar – but don’t throw out the analysis of patterns and rules
  • Get familiar with corpora – although frequency is not the only yardstick, the collocation and other patterns that corpora show us can be a useful supplement to our own intuitions about how words are used
  • Encourage the use of unanalysed chunks at an early stage
  • Avoid teaching many similar items together
  • Embrace translation: the most direct way to check the meaning of a chunk (really we are dealing with the comparison of chunks, rather than word-for-word translation)
  • Embrace repetition, drilling & other memory-based activities – how else will learners retain anything?
  • We still need a well thought-out lexical syllabus – or, if such a thing is impossible or undesirable, a syllabus which deals with lexical chunks as well as grammar patterns

Hoey, Michael. 2005. Lexical Priming: A New Theory of Words and Language.
Jordan, Geoff. 2013. “Lexical Syllabi and Lexical Chunks”.
—————–. 2014. “Hugh Dellar and the Lexical Approach”.
——————. 2014. “Hugh Dellar and the Lexical Approach Pt 2”.
Lewis, Michael. The Lexical Approach: The State of ELT and a Way Forward.
Nattinger, James R. and Jeanette S. DeCarrico. 1992. The Lexical Phrase and Language Teaching.
Pawley, Andrew and Frances Syder. 1983. ‘Two puzzles for linguistic theory: Nativelike selection and nativelike competence’, in J.C. Richards and R.W. Schmidt (eds), Language and Communication.
Willis, Dave. 1990. The Lexical Syllabus. (available free from
Woolard, George. 2013. Messaging: Beyond a Lexical Approach

Extracts from the British National Corpus via
See also this useful video from Scott Thornbury talking about COCA, the US English corpus:
Finally, a very easy webpage for checking collocations of a word:

See also the following blogs:


“The development of …”

Source: BNC

Source: BNC

What should not be done, or should one not do, “in isolation”?

Captura de pantalla 2015-02-08 a las 15.17.12

What do we typically “come across”?

Captura de pantalla 2014-11-26 a las 16.52.36

About the author : Neil McMillan

Current president of SLB and a freelance teacher, writer and teacher educator. Course director on the SLB TBLT course. Author and collaborating professor at the Universitat Oberta de Catalunya.

Leave A Comment

  1. mura nava 8 February 2015 at 10:02 pm - Reply

    hi Neil nice write-up and thanks for my blog shoutout :)

    regarding the limitations of relying too much on frequency of language use, this needs to be noted however i go along with those who make the case that other things being equal frequency of use is a pretty good proxy (though Geoff has pointed out that descriptions of language use cannot be used as measures for prescriptions in language teaching)

    it is interesting to note that the phrasal verb come across is not in the top 150 most frequent phrasal verbs (


  2. Neil McMillan 8 February 2015 at 11:14 pm - Reply

    haha oops … interesting although I can’t seem to see the whole list without paying for the original article. One thing I noticed which is maybe another argument against a frequency-based approach. The example you cite in your Google+ post was “look up”, which the PHAVE dictionary cites the following meaning as the most frequent:

    “Raise one’s eyes (88%) – He looked up from his book and shook his head.”

    For me however, and I was trained to think this way (although I’m not sure of the provenance of the advice any more), in this definition “look up” is not a phrasal verb – it is the verb “look” + adverb “up”. The meaning is therefore transparent, like with “go down” (the street) and “turn around”, and has the possibility (I’m guessing) of a more or less direct translation to other languages. And the PHAVE dictionary gives the literal meanings of go down and turn around as the most frequent, too. I don’t think these are the meanings that students tend to struggle with.

    Do you know Norbert Schmidt’s PHRASE list? It uses multi’word phrases which are semantically opaque. In the list relating to the first 1000 most frequent words we find “used to”, “so that” and “in order to”, phrases normally left to higher levels. Is an argument against a frequency-based syllabus perhaps that beginner and elementary students may not yet have enough other language to successfully incorporate subordinating structures like “so that” and “in order to”?

    One more thought about “come across” – in certain contexts I’m sure this phrase would appear more frequently than other phrasal verbs, for example in certain research fields and when reporting the history of research or ideas. But I guess any frequency list needs to be examined in terms of student need.

    • Neil McMillan 9 February 2015 at 12:52 pm - Reply

      Mura, thanks for the link to the list and I might have known you’d have been on to the PHRASE list quicker than a shot off a shovel!

  3. paul walsh 9 February 2015 at 12:02 pm - Reply

    Hi all,

    Cannot help but agree with everything that is said here – excellent stuff. The big problem with the lexical approach is, as you say, a lack of a coherent theory, methodology and methods. This means it cannot easily be ‘packaged’ for working teachers – in the same way as other methodologies. Perhaps we need to think of new learning activities? (I would love the days of the gap fill to be numbered…)

    One thing I might also add is the concept of ‘noticing’. I know that this is somewhat of a buzzword, but I think that learners need an environment where they have more cognitive ‘space’, and where they can conceptualise their own language learning more. When a classroom is full of coursebooks, IWBs, and all the other paraphernalia this is impossible.

  4. Neil McMillan 9 February 2015 at 1:00 pm - Reply

    Thanks Paul,

    I agree about noticing and admittedly we didn’t discuss this too much in the session. I think Geoff spotted a contradiction between Hoey’s concept of noticing (that it’s subconscious) and Lewis/Dellar’s (that it’s conscious). Isn’t Richard Schmidt the man on this issue? he talks about conscious noticing I think, but I guess the issue is far from settled.

    PS Let me apologise publicly for abandoning the TAW group for so long. I promise I’ll be back to catch up, with a string of excuses and a renewed vigour …

  5. Geoff Jordan 9 February 2015 at 8:16 pm - Reply

    Very impressive write-up of what sounds like an excellent session.

    My main question about “teaching lexically” is, simply: How do you do it? How do you “put more focus on lexical chunks”? How do you “encourage the use of unanalysed chunks at an early stage”?

    If you take Hoey’s uncritical endorsement of Krashen seriously then we subconsciously learn all the collocates, semantic associations, pragmatic associations and colligates of a word, plus textual collocation, textual semantic association, and textual colligation by being exposed to naturally occurring data that interests us and “slightly extends” us. In other words, no teaching is necessary. This view is, of course, contradicted by SLA findings. It’s also contradicted by Dellar who thinks we need to notice (in Schmidtt’s sense of the term), patterns in the input, but Dellar fails to say how as teachers facilitate noticing.

    There’s also the issue of precisely what lexis and which lexical chunks you focus on. Once you move away from the obvious criterion of frequency, what informs your decisions? For example, what’s the value of telling students (or getting them to guess, or whatever) that “skills” is the strongest collocate of “The development of..”, or that you’re more likely to come across a man than come across an interesting recipe for rice pudding?

    What seems to be a simple and compelling argument – English as a language is lexically driven – turns out to be a very slippery idea to get a pedagogical hold of, and those who rush in to prescribe teaching lexically are, in my opinion, more likely to be impulsive fools than timorous angels. See: the collocate for “timorous” is “beasties”, but the well-known saying requires that angels is the noun. :-) .

    • Neil McMillan 9 February 2015 at 11:10 pm - Reply

      Thanks for the comments Geoff. You’ve already pretty much convinced me of your case, I just have some thoughts about some of the points you raise.

      “How do you put more focus on lexical chunks?” – e.g. by highlighting them as they come up in input, encouraging students to notice and record them as chunks, to recycle them as chunks and to correct them as such when they are misused. If, as soon as we teach the verb “to depend” we associate it with “on”, drill “it depends on” + object, and correct “depends of …” whenever it occurs during accuracy-focused feedback – rather than leave it until the Upper Int book that suddenly mentions dependent prepositions (pun unintended), and presents 20 examples – I would hope that this would have some impact on learning. But I have absolutely no evidence for this, and the problem with action research on points like these, I’m never with a group long enough to come to any conclusions.

      As for the unanalysed chunks question, I feel this already happens in many cases with certain “elementary” structures like “have got” and “let’s go”. In the Scottish ESOL context I used to work in, we would teach “Have you got any ….?” and “I’ve got …. ” as chunks. It didn’t seem expedient to go into the mechanics of the present perfect and the slippery meaning of “any” (and I think expedience/efficiency does come into this somehow). Again, though, it’s hard to prove the benefits of this. I found, of course, many students who seemed to get stuck on “I got”, but it occurred to me that this was down, in some cases anyway, to a pronunciation issue and difficulty with the consonant cluster (like the Spanish students who know -ed endings, write -ed endings but find them very difficult to produce orally). On the other hand, informal American English might have influenced them. And on the third hand, my chunking may have served no purpose whatsoever.

      “The development of ____” example you mention is a bit pointless, I admit, although there’s definite value in focusing on the structure “the x of y”, insofar as (again Spanish) students tend to overapply the possessive “s” and come out with stuff like “skills’ development” (in this case, only an error you notice in writing, obviously). But for me the value of the corpora is simply being able to go to something which supplies more info than I can think of off the cuff, and decide which of it may be relevant to students’ needs at that time, or could help clarify meaning/use. When corpora demonstrate, for example, that “a bit” very frequently precedes negative concepts, or criticism, like “this reply is a bit longwinded”, then this kind of conceptual collocation (if I can call it that) can surely encourage, rather than inhibit, linguistic creativity.

      Your point about syllabi – would a process syllabus such as those you advocate elsewhere simply mean that the lexical component of that syllabus comprises those chunks which seem necessary to the fulfilment of communicative purpose or completion of tasks? The tasks themselves being determined via needs analyses, but the chunks being determined by the task requirements and the students’ negotiation of them? In other words, rather than a pre-constructed lexical syllabus, we would have a retrospectively constructed one which would vary according to the context, needs and level of the students, and in which the teacher has a significant responsibility in deciding the “seems necessary” part.