Try this, it works! 4….3….2…

Paul Nation‘s 4/3/2 1 activity is often cited as an excellent way to improve a students spoken fluency. I have used it myself and incorporated it into a lot of materials. It’s evidence based after all, right? 


The other day it suddenly hit me that I’d never once bothered to check if the method itself had any evidence to support it at all. Here’s what I found. 

The method 

In a paper from 1989 Nation describes the technique which originates in a 1983 paper by Maurice (if anyone has a copy I would love to read it). Ask students to prepare a talk on a given topic but don’t let them make any notes. Pair the learners up and give them 4 minutes to talk about it. Their partner should not interrupt. 

Next, the speaker and listener switch and the process is repeated. After this the students switch partners and repeat the process but this time with the time reduced to 3 minutes. In the final iteration the time is reduced to 2 minutes. The point of this is to get students speaking more fluently. 

In order to test the efficacy of this method Nation recorded students doing the activity and then measured various aspects of the performance. For example her measured, the number of words per minute, the number of hesitations and the number of repetitions. Nation also measured accuracy by counting the number of errors per minute. 


  • The 432 technique seemed to lead to improvements in student fluency as measured by words per minutes (WPM), and a reduction in hesitation and repetition. 
  • There was ‘some’ improvement in student accuracy. 
  • Students seemed to get better at only including important information (control of content). Nation believes this shows students may get better at condensing information. 

    Some thoughts

    The study was N=6. That’s very very small. Nation claims that despite this “the consistency of the results indicates that their gains from the activity would be typical of other learners” but with 6 advanced students can he really make these claims? 
    There seems to be a lot missing from the methodology section of this paper. What was classed as a mistake? how long was a hesitation and so on. There are no transcripts to look at to see what kind of language the students produced and we get no information about whether any students finished before the allocated time and if so how that was recorded. 

    Did students take it slowly on the first 4 minute round because they had some much time? would student have been able to speak at the “final” speed if they had only been given 2 minutes from the beginning? If the time wasn’t changed at all would students speed up anyway? 

    The “audience” students (those listening) do nothing other than listen. Whenever I do this activity I ask students to take notes to answer questions later on so they are at least paying attention. I also tend to start with a much shorter initial time. 4 minutes seems like a really really long time. The IELTS speaking test “long turn” only lasts for 2 minutes. 


    I was curious if anyone had replicated the study, perhaps with a larger cohort. It turned out they had. 


    Nation himself repeated the study 2 years later with Supot Arevart, only this time with 20 intermediate level students. This time, the authors give a bit more detail about the procedure. It seems that students were grouped in fours and spoke once, listening three times (which seems a bit dull for a speaking lesson). This paper also contains transcripts of participant speech. 


    • Again there was an increase in WPM (18 more words from the first to the last performance) though it is not clear if the repetition or time reduction is the cause.
    • Individual student results are listed in this paper and show that whereas some subjects made great gains (an increase of 48 WPM) others did not. One student actually got worse (though that was a student with an very high initial WPM count). 
    • Hesitations decreased by an average of around four fewer hesitations by the 3rd round. Again, the individual data shows us that the results were much more varied at an individual level with one participant going from 10 to 18 hesitation by the final round (incidentally the same student who did not make WPM gains).  

      De Jong and Perfetti (changing the topics)

      This paper tested 4/3/2 with a group who repeated the same topic and one that got a new topic every round. They found that even if the topics differed for each time the levels of fluency increased. This would seem to indicate that the reduction in time alone can prompt students to appear more fluent. However, the authors also found that only those who had repeated their topic showed improvement in the posttest. 

      Boers (keeping the time constant)

      Another two replications were carried out by Boers who is quite critical of Nation. He raises the following issues:
      • Nation claims that this is a well-researched technique (Nation and Newton 2009) when in actual fact there are very few studies into the technique. 
      • Nation research only shows improvements within the 4/3/2 activity itself. He does not show that this permanently affects a students fluency. 
      • Although Nation has claimed that 432 also improves accuracy and complexity, the actual results do not support that claim. 
      • Boers wonders if the repetition alone, without the time pressure, may have the same effect. 
      Boers studies (2014, N=10 and 2015, N=20 with Thai) introduced a control group with no time reduction (3/3/3 as well as 4/3/2)


      • As with Nation, Boers found increases in the number of words and WPM from the first to the third round. 
      • Both papers found the number of disfluencies (hesitations) decreased. 
      • They also found that in the time-constant group students improved as well, though not by as much. 
      • One striking finding was the amount of verbatim duplication in the shrinking time condition. This was notable to the authors as in some cases up to 50% of the texts were exact repetitions. 
      • “There was no compelling evidence of increased lexical sophistication and no evidence of increased syntactic complexity.”

      Boers found that participants seemed to repeating the same structures and this held for both replications. Although students got faster, errors did not decrease significantly and in some cases they increased. Boers notes that as students were just repeating the same thing without getting any input on, or evaluation of their performance, this is hardly surprising. 

      Take away

      Most of the research shows that repeated practice will increase fluency as measured by WPM (or syll/sec). However it will probably not do much for accuracy or complexity. As Boers argues that “the 4/3/2 technique is recommendable if the sole aim of the activity is to push fluency”. But you can increase fluency merely by reducing the time students have, even with different topics. You can also get a decent increase with no reduction in time. 

      Simply put, rehearsing a talk and repeating it will tend to improve fluency. A modified 432 in which students get feedback on their performance (from other students or the teacher) and then repeat it may help with accuracy and complexity. But as with many things “more research is needed”. 

      2 thoughts on “Try this, it works! 4….3….2…

      1. No comments on this one? How come? I'll put it down to a post-Christmas slump. It's a topic I have been looking into myself lately.Everyone who writes about this technique cites Maurice, but I don't think many of them have actually seen the source! Maurice's copy is impossible to find – so can't help you there, I'm afraid. Anyhow, in the recent – and, by some accounts, rather disappointing book – co-authored by Paul Nation (with Stuart Webb) entitled How Vocabulary is Learned they mention that the original technique, as suggested by Maurice, was in fact 5-4-3 (but everybody refers to it as 4-3-2 and cites Maurice). It was apparently Nation who modified it to 4-3-2 because of a lower level of proficiency of the learners he describes in his 1989 study, which you mention above.Even more strikingly, the authors go on to misinterpret Boers's conclusion. In summarising Boers's 2014 study, they claim that 'the decreasing time was the main contributor to fluency development, not repetition'. But according to Boers, it's exactly the opposite!aLeo


      2. Hi Leo, Thanks for the comment. Most of my posts don't actually get many comments -not sure how normal this is. Someone (not sure he wants to be named) sent me the original article shortly after I posted this. It wasn't originally 5/4/3 in the version I was sent.


      Leave a Reply

      Fill in your details below or click an icon to log in: Logo

      You are commenting using your account. Log Out /  Change )

      Google photo

      You are commenting using your Google account. Log Out /  Change )

      Twitter picture

      You are commenting using your Twitter account. Log Out /  Change )

      Facebook photo

      You are commenting using your Facebook account. Log Out /  Change )

      Connecting to %s