Implicit Learning and Input Enhancement – Long (2017) Part 2

The second of two posts on instructed second language acquisition (ISLA) investigates this new subfield at the intersection of second language acquisition (SLA) research and language teaching (Long 2017). In the first post we focused on his definition of ISLA and how it differs from L2 teaching and classroom research. Now we move on to the second part of the article to understand why implicit language learning is favoured by second language researchers and what this might mean for teachers.

A cognitivist-interactionist theory of ISLA

Adults learning an L2 are disadvantaged because a) first language learning has tuned out the capacity we have as newborns to learn any language (i.e., notice all kinds of phonetic contrasts or grammatical features) and b) after age 12, our capacity for implicit learning is much reduced. These factors block L2 users’ capacity to notice L2 features, creating a learning problem.

ISLA research aims to determine what kind of instruction can help. Long assumes three constraints on L2 instruction.

  1. Explicit or implicit learning alone is not sufficient: there is too much to learn.
    EFL learners need to know 9K word families. Class time cannot cover this volume of vocabulary, and for incidental learning via reading a 3 million word corpus would be required, also too time-consuming.
  2. Instruction can only affect the linguistic environment, NOT the actual learner.
    Implicit and incidental learning can occur alongside explicit knowledge gained from intentional learning. It is difficult to know what kind of learning results from instruction, although it is still possible to design tasks to encourage one type or another by focusing on language form or communicative intent.
  3. Implicit L2 learning is more important for a functional command of L2 than explicit learning, and some L2 researchers do not believe that explicit knowledge can become implicit (the “non-interface position”). This is still a minority position in language teaching, and perhaps rightly so, since few learners have both time and aptitude for implicit language learning, Long acknowledges.

Long suggests three ways instruction can support L2 acquisition given these constraints. All rely on incidental learning plus one of the following:

  1. focus on forms to facilitate noticing
    Teach grammar explicitly, by traditional presentation-practice-production (PPP), to make learners aware of L2 features.
  2. focus on form to facilitate noticing or detection
    Include brief references to L2 features during communicative teaching to make learners notice, or become conscious of a feature. Later, learners may apply this learning subconsciously, that is, detect or learn from further examples in subsequent input.
  3. unintrusive input enhancement to facilitate detection
    Develop unconscious awareness of L2 cues, or make L2 features salient to learners without explicit noticing, by highlighting important features like verbal morphology with bold or colour, or by associating visual and aural input.

Long argues that Condition 1 is effective only if L2 competence is tested on discrete points and without time pressure. In other words, it doesn’t produce implicit learning, and it disrupts communicative teaching. Condition 2, drawing learners’ attention briefly to particular L2 features during communicative activities, also works. This has the advantage of not sabotaging communicative instruction – see Geoff Jordan and Anthony Schmidt for more on this.

Condition 3 would have even less negative impact on communicative instruction: the goal is to to reset L1 parameters or “unlearn” instincts derived from L1 by focusing on relevant L2 features. Cintrón-Valentín & Ellis (2015) investigated the learning of L2 Latin verbal morphology by L1 English speakers. L1-influenced reliance on adverbial cues blocks learners’ attention to more important verbal morphology. The study found that highlighting morphological cues through colour coding in computer-delivered input increased attention to this salient feature and helped learners’ overcome the adverbial block to improve learning. This suggests more explicit grammar teaching is not necessary.

Condition 3 can also include ways to speed up detection of new forms in input. Malone (2016) investigated incidental vocabulary learning through reading with and without an oral rendition. He found the bi-modal condition (reading and hearing) produced better learning. He suggests teachers recommend learners read along with audiobooks to improve incidental learning. Long claims that basic research has shown that unintrusive input enhancement works, and that controlled laboratory studies should now be undertaken.

An ISLA research agenda

In conclusion, Long rejects explicit grammatical instruction such as Ellis’ hybrid grammatical and task syllabus, or a grammar syllabus supplemented by extensive reading, claiming they lack psycholinguistic coherence. He maintains his preference for incidental learning via a communicative syllabus with opportunities for focus on form, or, better, unintrusive enhancements to allow unconscious re-tuning of attention to L2 features and uptake of new forms.

Long sees a major goal of ISLA research as freeing instruction of unnecessary artificial aids, to identify the least intrusive but still efficient means of achieving instructional goals.


I like how this article carves out a space at the intersection of second language acquisition and language teaching research. SLA research is not all relevant to language teaching, since some relates to naturalistic learning, for example, or consists in basic research without direct classroom applications. Similarly, Long shows how not all research in language teaching or L2 classrooms relates to language acquisition or development. This speaks perhaps to some of Geoff Jordan’s concerns about the sociocultural turn in applied linguistics, and to Richard Smith’s points about teacher research: while not denying the interest of this work, Long is clear that a) very specific pedagogical choices are best made by teachers in their own classroom contexts, and b) the context-dependent findings of research into pedagogical procedures “do not qualify as research on ISLA.”

Since the article focuses on research rather than instruction, the arguments are framed around learning rather than teaching, and Long does acknowledge that SLA researchers’ emphasis on “incidental and implicit L2 learning in adults” is “still a minority position in the world of language teaching” (p. 23). He argues that explicit teaching should not take the form of a grammatical syllabus, but rather focus on form, or unintrusive input enhancement. His examples are based on computer-delivered instructional materials, making the teacher at best an instructional designer. Focus on form, however, also concerns teacher behaviour in the classroom, including feedback on errors, for example. There is no mention in this article of how unintrusive input enhancements might be provided and how, or indeed whether, research into this could contribute to ISLA. Thus for the moment, it seems that ISLA remains more focused on learning than teaching, no doubt to the frustration of some readers of ELT Research Bites.

Cintrón-Valentín, M. and Ellis, N. (2015) Exploring the interface: explicit focus-on- form instruction and learned attentional biases in L2 Latin. Language Learning 37(2): 197–235.

Long, M. (2017). Instructed second language acquisition (ISLA): geopolitics, methodological issues, and some major research questions. ISLA, 1(1): 7-44. PDF

Malone, J. (2016) Incidental vocabulary learning in SLA: effects of frequency and aural enhancement (Qualifying Paper. PhD in SLA Program). College Park: University of Maryland.

Featured image by Geralt

Shona Whyte on TwitterShona Whyte on Wordpress
Shona Whyte
I’m Shona Whyte. I’m an associate professor of English at the Université Nice Sophia Antipolis in Nice, France. I teach EFL, TEFL and CALL and my research focuses on technology for teaching and learning languages.

I blog now and again on second language teaching and research, and curate resources on Teacher Education in Languages with Technology at

5 thoughts on “Implicit Learning and Input Enhancement – Long (2017) Part 2”

  1. Hello Kyle – thanks for your comment: you’re right, these terms are worth examining further. There is a big literature on the implicit/explicit distinction and an(y) interface between the two forms of learning. You could start with the Ellis’, or I have this little overview here

    Ellis, N. C. (2011). Implicit and explicit SLA and their interface. Implicit and explicit language learning, 35-47.
    Ellis, R. (2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies in second language acquisition, 27(2), 141-172.

    1. Thank you for your reply, Shona – Ellis’s 2005 paper is particularly useful for understanding the different theories and conceptions of types of knowledge and the interface between them.


      1. Yes. The Long (2017) paper references more recent work by the other Ellis, though as I argue in my post the bulk of the work being done on implicit learning involved computer-delivered materials, not classroom practice.

        Long himself claims there is never enough instructional time for teachers to aim for only implicit learning, so some explicit learning should be involved too. The question is simply how to do this efficiently, without disrupting any implicit learning.

  2. Hi Shona,

    You’ve mentioned ‘explicit knowledge’ a couple of times in this post and also stated the following in a previous post: “All second language researchers interested in psycholinguistics assume a distinction between a deliberate, consciously controlled type of linguistic knowledge (‘explicit’ knowledge) and an intuitive, automatic type of linguistic knowledge (‘acquired’ or ‘implicit’ knowledge).” (

    The assumption that there are in fact two separate types of ‘knowledge’ seems to require more attention than it typically receives. Before discussing the interface between them, whether one type is better or how we can best develop either type, I think it would be appropriate to first discuss the evidence that they even exist in the first place. Instead, it seems to me that a lot of SLA research and discourse proceeds without doing so and we should, of course, always be wary when things proceed on the basis of an unchallenged assumption.

    Perhaps this would be best addressed in a separate post but my question is: what empirical evidence is there to support the hypotheses that these two ‘knowledges’ exist?


Leave a Reply