Skip to main content

Fluency development in language learning

This post is a summary of a chapter in the Loewen and Sato book The Routledge Handbook of Instructed Second Language Acquisition (2017). The chapter was written by Tracey M. Derwing from the University of Alberta. I've previously dipped into this book for summaries of research about vocabulary and grammar. The topic of fluency is of interest to me at the moment as Gianfranco and I are doing our research before publishing a book next year (provisional title Acquiring Second Language Skills: From Comprehension to Fluency).

I'll add my own glosses to this summary as I go along (in italics).

Derwing begins her chapter with a quotation about fluency from Charles Fillmore, who described one of four types of fluency as the "ability to talk at length with few pauses" (Fillmore, 2000, p.51). 

(This is what many of us think when we describe what it means to be a fluent, although we may also just mean "proficient" in the language. We have an idea that someone can speak in a way which is like a native speaker, or at least approximates to that. To be "fluent in French" is to be able to speak quite fast, without too much hesitation and probably with some accuracy.)

Derwing goes on to point out that research into fluency does, in fact, tend to focus on temporal measures of L2 production - syllables per second, number and length of pauses and mean "length of run" - number of syllables between pauses, not general proficiency. It's about how well speech flows.

However, as Segalowitz (2010) has explained, when looked at from a psycholinguistic perspective, there are actually three types of fluency:

  1. Cognitive fluency: our ability to mobilise and bring together the underlying cognitive processes responsible for producing fluent utterances - short-term memory, planning, lexical retrieval, choice of grammar, phonological knowledge and so on.
  2. Utterance fluency: this is what I referred to above, namely how well speech flows. This is like the oral manifestation of cognitive fluency.
  3. Perceived fluency: judgments made about speakers based on samples of their speech.

(I have sometimes thought that the third of these, perceived fluency, may be of academic interest, but is less useful from a teacher's point of view. The teacher may be keener to know what the sources and manifestations of fluency are and how to help them develop, rather than perceptions of fluency.)

Derwing soon makes the point that "both massive amounts of input and opportunities to speak are necessary to improve L2 fluency". In addition, being more fluent encourages learners to engage in more conversations which produce more input.

(In passing, I'd note two things here. Firstly, the "opportunities to speak" part is significant and may run counter to some views about language acquisition which hold that all you really need is input. Secondly, while it is generally true that masses of input is needed to produce ultimate fluency, you can be fluent on a limited repertoire of language - more of that later.)

Fluency is of interest to researchers in various fields, but applied linguists (and teachers) are often interested in identifying "ways in which learners' fluency can be enhanced through manipulation of tasks in the classroom, the effects of study abroad and other forms of immersion..."

Drawing on research into L1 acquisition, Derwing cites Pawley and Snyder (2000) who came up with the 'one-clause-at-time' constraint. This means that speakers are limited by their memory capacity, so can only focus on a single clause at a time. The writers stated: "it is the knowledge of conventional expressions, more than anything, that gives speakers the means to escape from the one-clause-at-a-time constraint and that is the key to native-like fluency" (p.164).

(This is the point that we very often call upon ready-made chunks of language and chain them together to be fluent. Even near-beginners are able to do this to an extent, once they have a stock of phrases to call upon. It's one of the reasons why it is recommended to place a great focus on chunking in lesson planning.)

Derwing cites a six-month long study by Wood (2006) which showed the extent to which the use of formulaic sequences contributes to fluency. But other aspects of cognitive processing are also important. Following de Bot (1992) you need to start with an idea (a semantic notion) that you want to express (the 'conceptualiser', which includes the choice of which language to use). You then need to put the idea into linguistic form (the 'formulator'), where words, grammar and phonology are implemented. De Bot suggests that for L2 learners you need a 'feedback loop', as you realise that you don't have the words you need and have to reformulate the language to match what you can actually do. 

The next stage is labelled the 'articulator', the component in which speech is actually produced. As articulation proceeds, you monitor what you are saying for accuracy. In this model, word finding speed is crucial.

Segalowitz (2010) adapted the de Bot model, identifying poinst where fluency can break down, e.g. in encoding of grammar, lexical retrieval, phonological encoding, articulation and speaker's perceptions of their own production. The degree of automaticity a speaker possesses dictates fluency. Fluent speakers perform the process of speaking with little effort. Learners have to resort to paraphrase or even conceptualise what they want to say based on their limited range of language.

(Clearly, the more words and chunks of language learners have in long-term memory, the easier it is both to retrieve and to paraphrase. In addition, articulation is easier if students have mastered a good level of pronunciation skill.)

Derwing goes on to look at empirical studies of fluency. For example, Towell, Hawkins and Bazergui (1996) found that there was a correlation between speech rate in L1 and L2. In general, if someone speaks fast in their first language, they are more likely to speak fast in the second. In another study by Derwing et al (2009), it was found that fluency in the L2 was influenced by the first language of the learners. In this case, L1 Mandarin speakers became less fluent in English than L1 Slavic language speakers. So it may be the case that where there is a greater dissimilarity between L1 and L2, fluency is slower to develop. Derwing points out, however, that other factors may have been at play in the study quoted - for example the amount of English the Slavic speakers engaged with overall, e.g. online.

(This, on the face of it, makes sense, since where two languages are similar lexically and/or grammatically, it is easier to draw on the L1 vocab and grammar knowledge to string chunks together. In general, it is easier to learn a language similar to one's first.)

Derwing next explains how the choice of classroom task affects fluency. Derwing et al (2004) compared three tasks: a picture narrative, a monologue and a conversation. The picture narrative was judged by listeners to be less fluent than the other two conditions. This was attributed to the fact that the picture tasks dictated to some extent which vocabulary could be used. In the monologue and conversation speakers could choose their own language.

Foster and Skehan (1996) found the same, and attributed the problem with the picture task to greater cognitive load - but essentially they are referring to the same problem - one of free choice.

(This has implications for test design of course. At GCSE, if you allow students to present or hold conversations, they are likely to be more fluent than when having to do a role play of picture task, where the language is somewhat prescribed.)

The ability to plan ahead of a task helps fluency (Ellis, 2009). In general, the longer you have to plan, the better. It remains unclear to what extent it helps to guide students with the planning, or just let them do it themselves.

(You might surmise that the less sophisticated and fluent learner would benefit from the most guidance, e.g. getting students to think about what strategies to use when preparing to do a picture task or role play.)

When it comes to the pedagogical implications, Derwing makes the following points.

Pedagogical activities can increase automaticity through increased awareness of planning and rehearsal tasks. Teaching frequently occurring chunks, common discourse markers (e.g. words and phrases like then, next, after all, finally). Crucially, "an intensified focus on speaking and listening tasks".

(It's the old adage - use the language, don't talk about it. If you provide lots of comprehensible input and opportunities to interact with it, fluency will gradually develop.)

Factors which may militate against fluency practice include large class sizes, competing demands of other language skills, time limitations and lack of teacher familiarity with activities which promote fluency.

(I might add lack of teacher fluency - usually no fault of the teacher - and syllabus requirements, which may, for example, require that focus on other skills mentioned above. Teacher beliefs also come into play, for example the belief that spending a good deal of time teaching the rules of grammar and conjugations will lead to fluency.)

Practising outside the classroom is also important, as Derwing notes.

As regards specific lesson ideas, Derwing quotes Nation's (1989)  famous 4,3,2 task whereby students have to retell the same story in progressively shorter lengths of time. nation found that successive tellings of the story resulted in fewer pauses and fillers.

A model of fluency development by Gatbonton and Segalowitz (2005) is then described, perhaps most relevant in the context of task-based language teaching. They call it ACCESS (Automatisation in Communicative Contexts of Essential Speech Segments" (!). This involves a pre-task, a genuinely communicative, inherently repetitive group task and form-focused instruction where needed,

(This reads pretty much like the type of communicative task familiar since the late 1970s, e.g. in the work of Penny Ur.)

Derwing looks at the work of Nation and Newton (2009) who argue that you need a focus on all the language skills for fluency to develop. nation and newton suggest that tasks centred on fluency should be based on language with which students are already very familiar. The emphasis should be on communication under time pressure.

(This brings me back to a point I made earlier - you can practise fluency on a limited diet of language. Within Gianfranco Conti's MARS-EARS framework, the fluency practice would occur late in the sequence, under, I suggest, the R for Routinisation and S for Spontaneity. You can anticipate that when we write bout fluency we shall suggest a good range of specific classroom tasks with fluency building in mind.)

Reference is next made to Rossiter et al (2010) who outline a good range of classroom activities.

A study of galante and Thomson (2016) is also described. It compared two groups of Brazilian learners of English. One group were taught over four moths (74 hours) using communicative methods with pair work etc, while the other used drama and role play methods. In this study, the group using drama appeared to gain more fluency.

(I wonder what other factors would have been in play in this study, e.g. the role of the teacher.)

Finally, I'll summarise some of the teacher tips Derwing mentions:

  • Raise students' awareness of markers of fluency such as appropriate intonation to show the speaker still holds the floor; placement of pauses or clause boundaries; explicit instruction of oral fluency (I'm not sure quite what that means here).
  • Have students transcribe YouTube videos which the class can analyse together.
  • Use 'shadowing' where learners read along with a transcipt, echoing the speaker's pronunciation and intonation.
  • Use rehearsal and repetition tasks (e.g. the 4,3,2 task).
  • Explicitly teach high-frequency chunks.
  • Teach fillers such as in English  "like", "you know", "first".
  • Create role play tasks.

(The range of classroom ideas in the chapter is disappointingly short, alas.)

Derwing then goes on to look at study abroad settings, which I'll skip over here, then future directions for research.


de Bot (1992). A bilingual production model: Levelt's speaking model adapted. Applied Linguistics, 13, 1-24.

Derwing, T.M.& Munro, M.J. (2013). The development of L2 oral language skills in two L1 groups. Language Learning, 63, 163-195.

Derwing. T.M., Munro, M.J. and Thomson, R.I. (2008). A longitudinal study of ESL learners fluency and comprehensibility development. Applied Linguistics, 29, 359-380.

Derwing. T.M., Munro, M.J. and Rossiter, M.J. (2009). The relationship between L1 fluency and L2 fluency development. Studies in Second language Acquisition, 31, 533-557.

Ellis, R. (2009). The differential effects of three types of task planning on the fluency, complexity and accuracy in L2 oral production. Applied Linguistics, 30, 474-509.

Fillmore, C.J. (2000). On fluency. In H. Riggenbach (ed.) Perspectives on fluency. University of Michigan.

Foster, P. and Skehan, P. (1996). The influence of planning and task type of second language performance. Studies in Second language Acquisition, 18, 299-323.

Galante,. A. & Thomson, R.I. (2016). The effectiveness of drama as an instructional approach for the development of L2 fluency, comprehensibility, and accentedness. TESOL Quarterly.

Gatbonton, E. & Segalowitz, N. (2005). Rethinking communicative language teaching; A focus on access to fluency. Canadian Modern language Review, 61, 325-353.

Loewen and Sato (2017) The Routledge Handbook of Instructed Second Language Acquisition. Routledge.

Nation, I.S.P. (1989). Improving speaking fluency. System, 17, 377-384.

Nation, I.S.P. 7 Newton, J. (2009). Teaching ESL/EFL listening and speaking. Routledge.

Pawley, A. & Syder, F.H. (2000). The one-clause-at-a-time hypothesis. In H. Riggenbach (ed.) Perspectives on fluency. University of Michigan.

Rossiter, M.J., Derwing, T.M., Manimtim, L.G. & Thomson, R.I (2010). Oral fluency: The neglected component in the communicative language classroom. Canadian Modern Language Review, 66, 583-606.

Segalowitz, N. (2010). The cognitive basis of second language fluency. Routledge.

Segalowitz, N. (2013). Fluency. In P. Robinson (ed.). The Routledge Encyclopedia of second language acquisition. Routledge.

Temple, L. (20056). Instructed learners' fluency and implicit./explicit language processes. In M. Pierrard & A, Housen (eds.) Investigations in instructed and second language acquisition. De Gruyter Mouton.

Towell, R., Hawkins, R. & Bazergui, N. (1996). The development of fluency in advanced learners of French. Applied Linguistics, 17, 84-119.

Wood, D.C. (2006). Uses and functions of formulaic sequences in second language speech. Canadian Modern Language Review, 63, 13-33.


Popular posts from this blog

What is the natural order hypothesis?

The natural order hypothesis states that all learners acquire the grammatical structures of a language in roughly the same order. This applies to both first and second language acquisition. This order is not dependent on the ease with which a particular language feature can be taught; in English, some features, such as third-person "-s" ("he runs") are easy to teach in a classroom setting, but are not typically fully acquired until the later stages of language acquisition. The hypothesis was based on morpheme studies by Heidi Dulay and Marina Burt, which found that certain morphemes were predictably learned before others during the course of second language acquisition. The hypothesis was picked up by Stephen Krashen who incorporated it in his very well known input model of second language learning. Furthermore, according to the natural order hypothesis, the order of acquisition remains the same regardless of the teacher's explicit instruction; in other words,

What is "Input Processing"?

Input Processing (IP) was proposed by Bill VanPatten, Professor of Spanish and Second Language Acquisition from Michigan State University. Bill may be known to some of you from his podcast show Tea with BVP. He is one of those rare university academics who makes a specific effort to engage with practising teachers. IP was first proposed in a 1993 article (published with T. Cadierno in the Modern Language Journal) entitled "Input processing and second language acquisition: A role for instruction." My summary of it is based on an article "Input Processing and Processing Instruction: Definitions and Issues" (2013) by Hossein Hashemnezhad. IP is a little complicated to explain, but I'll do my best to summarise the key points before suggesting how it relates to other ways of looking at classroom language teaching. Is this actually any use to teachers? I apologise in advance for over-simplifying or misunderstanding. To paraphrase Dr Leonard McCoy from Star Trek &q

Pros and cons of pair and group work

Most teachers have made frequent use of pair and group work for many years, notably since the rise of communicative language teaching in the 1980s. Even before then it would have been common for pupils to work in pairs on simple role-play and dialogue tasks. So pair and group work is standard practice, if not universally supported by language teachers. It’s always worth evaluating, however, whether a practice works - whether, in this case, it helps students develop their proficiency. Pros Rod Ellis (2005) summarises the advantages of pair/group work (based on Jacobs, 1998) “1. The quantity of learner speech can increase. In teacher-fronted classrooms, the teacher typically speaks 80% of the time; in groupwork more students talk for more of the time. 2. The variety of speech acts can increase. In teacher-fronted classrooms, students are cast in a responsive role, but in groupwork they can perform a wide range of roles, including those involved in the negotiation of meaning. 3. There can

New MFL GCSE consultation

Updated on 7th April, with a few modifications to the original post written about a month earlier. ........................................................................... The DfE in England has recently published information about the proposed new GCSE exams, first teaching September 2023, first exams June 2025. There are two consultations going on, one regarding the subject content, and the other (much shorter) with respect to the assessment arrangements such as tiering.  The context is important here. DfE are worried about uptake in GCSE MFL, especially with their EBacc target of 90% uptake in mind. (This is highly unlikely to be achieved.) Therefore they would like an exam which makes the subject more attractive, both in terms of interesting content and accessibility (how easy it is thought to be). They are aware also of criticisms levelled at current papers that the exam is elitist, featuring too much subject matter which appeals to middle class students. Recall that MFL has be

An NCELP lesson resource analysed

NCELP (National Centre for Excellence for Language Pedagogy) is the body set up and financed by the DfE in England. based at the University of York and headed by Emma Marsden and Rachel Hawkes. It works through a number of hub secondary schools which, in turn, work with a small group of other schools. Their mission is, broadly speaking, to spread the research findings and principles as laid out in the Teaching Schools Council (TSC) Review of MFL Pedagogy from 2016. By sharing a selected body of research, considered relevant to secondary MFL in England, and creating schemes of work and lesson resources across the hub schools, they hope to spread so-called best practice around the country. As I write this, schemes of learning and lesson resources have been written up to the third term of Y8 for French, German and Spanish. I've been watching with interest as these resources have been built up and in general my view has been that the research resources are very useful and informative (