Fluency, Accuracy and Scott Thornbury

Over the last three months I’ve gone through a process which has shredded to bits my grounding in teaching, thrown it up in the air and reformulated some of it, and shown me how to throw out the old and bring in the new. It has helped me to develop an approach to educating which is principled and grounded in sound pedagogic methodology. You’ve guessed it: the DELTA. DELTA Series Image

A topic which has cropped up again and again during the DELTA is the accuracy vs fluency dichotomy. Which of the two should there be more focus on? Which leads to better language users?

The accuracy camp of the discussion was born a long time ago, when language teaching was in its early days. This was the time when errors were feared, even avoided like the plague, and accuracy was the single measurement of a good language user.

As the communicative method came along, it became apparent that the number one use learners have for language is communication. Communication is centered around a communicative message, so the focus should be on conveying that message. That led to the natural conclusion: fluency beats accuracy in a communicative approach.

However, as Scott Thornbury pointed out in his workshop at Italy’s National TESOL conference 2014 in Rome, when you focus solely on fluency, then accuracy suffers greatly.

We know from history that a focus on accuracy leads to poor fluency, and an over focus on fluency leads to poor accuracy. So, where does the answer lie? Somewhere in the middle?

Scott pointed out that the “status quo” is to put the ball firmly in the accuracy camp at the beginning – when learners are around A1/A2 level – and it sways towards the fluency side as the learners work their way up the CEFR framework. However, he went on to suggest that it might be much more effective and more representative of First Language Acquisition if the focus were on fluency at the beginning and then later on accuracy. This echoes the idea of learning some vocabulary of a language, so that when you go on holiday you can get across your message, although it might be grammatically incorrect, such as saying to a Spanish barman “un cerveza” instead of “querí una cerveza.”

The current status quo would suggest that by the time a learner reaches C1 level, classes would be predominately focused on fluency, and our teaching experience confirms this: in higher level courses a number of lessons can pass by until a worthwhile grammar point arises – and even then it is probably more likely to be grammaticalised lexis than a pure grammatical form.

So, if we know a balance is necessary and we know how to measure accuracy (by the number of mistakes), then how do we measure and, more importantly, teach fluency?

This was the very question Scott Thornbury raised in his talk in Rome. He showed two short clips of young Italians speaking in English and he asked the audience to judge their fluency. To no surprise, it quickly became clear that the audience, who consisted mainly of non-native speakers, found one of the two speakers to be significantly more ’fluent.’

Scott enquired as to why and under what criteria they had measured this perceived ’fluency.’ The audience offered up the usual suspects: varied grammatical constructions, a wide range of lexis and slow, clear annunciation with little pausing.

I was quick to disagree and state I found the other speaker significantly more fluent. Several of the audience members highlighted that the other person in the video clip couldn’t possibly be fluent, because he:

  • Limited his grammar to simple structures
  • Had a poor range of lexis and repeated words and phrases very often
  • Paused a lot
  • Used a number of fillers, such as sounds like ”mmmmh” and repeating final words
  • Suffered a number of false starts, having to go back and fix his utterances by starting again

My response, which Scott agreed with, was that these are the exact criteria which measure fluency – these are exactly what native speakers do when speaking fluently and, in this particular example, the speaker did all of these and more, therefore he was very fluent.

This seemed to shock a good number of people but my assertion was backed up by other native speakers in the room, who quickly pointed out that they found this speaker much easier to understand because he did all the things natives do when they speak e.g. use simplified grammatical structures, pause, use fillers and false starts.

However, if there was any more reason to doubt the speaker’s fluency, the proof was in the pudding: Scott pulled up a transcript of the two speakers, which showed the first had a lot of mistakes, many of which seemed quite fossilized, and the second speaker (the truly more fluent speaker) had hardly any errors.

So, the speaker in this video clip was not only ’fluent’ in the same way native speakers are fluent, but he was also very accurate.

This led me to an idea I had during the DELTA a couple of weeks earlier: in order to be fluent, don’t we need to be accurate?

This is exactly what Scott went on to state, pulling up a slide which visually represented the accuracy-fluency dichotomy, as shown below:

accuracy fluency venn

Scott went on to look at a number of definitions of fluency until a consensus was reached:

“Fluency is the ability to convey the impression of idiomatic intelligibility in real-time”

In short, in order to sound ‘fluent’, you have to:

  • Use language structures appropriate to the speech act – more often than not this is simple structures
  • Use lexis appropriate to the topic – if you’re an Italian speaker and you use a lot of latinate words, you’re probably not going to give off the impression of being fluent, as English speakers prefer germanic-based words when talking informally
  • Use runs of words with pausing which reflects what natives would do.
  • Fill pauses with fillers
  • Use idiomatic language, repeat words and structures.

That brings us on to the question of how do you teach fluency in the classroom? Which exercises can you use? Do you have any suggestions or ideas? Please feel free to leave them below as a comment.

18 thoughts on “Fluency, Accuracy and Scott Thornbury

  1. Interesting~
    I think the example of Italian speakers is better to be used to demonstrate the level of authenticity, instead of fluency and accuracy.
    For example, a native speaker may not strike a high mark in TOEFL speaking test, because his speech contains a lot of ‘uh’, ‘mmmh’ , repetitions and grammar mistakes. This is considered by ETS to be an incompetence in fluency and accuracy, but actually, his speech is authentic.
    That’s why I believe what cannot be assessed is not fluency nor accuracy, rather than authenticity, since authenticity derives from personal experience and perception of the world, which is hard to be measured or quantified.

    Like

  2. Anthony, thanks for the interesting and stimulating post.

    One thing that I think gets lost in the fluency/accuracy discussion is how a student/teacher best goes about promoting both. Fluency, through meaningful input that is at the right level (CI) and accuracy through self correcting and noticing/monitoring. Too often we think of fluency as just speaking lots and accuracy repeating and practicing grammar points/use. This is much too behaviorist but makes the teacher seem like they are teaching and the students seem like they are learning.

    As to Mia’s question, I’d add to your list “voice”, one of the traits of writing that immediately give one the sense that the speaker is fluent and means what he/she says. This trait goes for speech also and is what you sensed in the speaker of Scott’s example.

    Like

  3. This is very interesting but seems to focus on spoken language. How would you go about measuring fluency in writing (or reading for that matter)? Accuracy is is easy to measure in writing, but fluency, not so much. I’m trying to put together a rubric for a base line writing assessment and I want to balance the results of the accuracy measure with a fluency measure. Kinda outta ideas…

    Like

    1. Hi! Thank you for the comment.

      That is a really good question! I guess in the talk/post the features of what make speaking fluent were outlined and then used to measure a learner’s spoken fluency. Maybe the same could be done for writing? For example: take a look at an authentic piece of writing (such as an e-mail to a work colleague, a short chat in a set of SMS messages, comments on Facebook or even on a blog) and decide what features make that authentic writing more or less fluent. I haven’t put much thought into it but some of the first things that come to mind are:

      – Logical ordering of ideas
      – Connection of ideas
      – Referencing: forwards, backwards and externally
      – Chunks or collocations
      – Adhering to the features of the genre: does the letter look like a letter? Does the comment look like a comment?

      Hope that helps. Have you finished compiling your list? Do you have a blog where you’ll be sharing it – it would be great to read.

      Like

  4. Thanks a lot for sharing this experience and what you drew from it, I wish I could hear the two speakers to get a clearer idea.
    I suppose that adult speakers feel confident (and thus become fluent) in speaking only when they have enough accuracy to be sure they are not making too many mistakes – Italians are often afraid to “lose face”!
    I often have a very hard time getting my A2 adult students to talk as they are afraid to make mistakes. That’s where accuracy-oriented activities might also help fluency: once they feel confident enough in a controlled activity, they might be less afraid to use the same language in a freer way.

    Like

  5. Well said. I have always wondered how teachers can ignore accuracy, especially with lower-level students. Understanding and using the right grammar, vocabulary, etc. prove so important, and then let’s them better understand and break down more challenging structures later.

    This is the first time to visit your blog, and I will definitely be back for more.

    Like

    1. Hi Chris! Thank you for visiting the blog! I am glad you are enjoying it 🙂 As I say in the About section of the blog (https://ashowski.wordpress.com/contact/) the whole idea is to spread as much knowledge as possible among English teachers around the world. If you ever feel like commenting on anything to share some ideas or thoughts, please don’t hesitate.

      Like

    1. Interesting~
      I think the example of Italian speakers is better to be used to demonstrate the level of authenticity, instead of fluency and accuracy.
      For example, a native speaker may not strike a high mark in TOEFL speaking test, because his speech contains a lot of ‘uh’, ‘mmmh’ , repetitions and grammar mistakes. This is considered by ETS to be an incompetence in fluency and accuracy, but actually, his speech is authentic.
      That’s why I believe what cannot be assessed is not fluency nor accuracy, rather than authenticity, since authenticity derives from personal experience and perception of the world, which is hard to be measured or quantified.

      Like

  6. Sounds like it was a fascinating session! 🙂

    It’s good to hear that Scott Thornbury places more importance on accuracy at the beginner levels such as A1-A2 and then encourages a shift to fluency in higher levels. To me it seems logical and that’s why I focus on grammar structures and controlled practice quite a bit with my A1-A2 group. Nice to feel like I’m on the right track with them.

    Like

    1. I think you may have that the wrong way round- at least regarding what has been said in the above post. According to this article, Scott Thornbury suggests that an emphasis on fluency over accuracy more closely mirrors L1 acquisition and therefore might be more effective. Key word being ‘might’… what does the research say?

      Like

      1. You’re absolutely right, Joey. Thanks for pointing that out. I was in a rush when I read the post and responded. I never seem to have enough time! 🙂

        I may need to rethink my focus and methods with my A1-A2 class. Food for thought. Thanks.

        Like

Tell us what you think...

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s