humanizing pt2

Part 1:  Knewton and its role in ELT
Part 2:  Open platforms and teacher-driven adaptive learning

[Note: we’ll be presenting a case study on how we’ve put these ideas into practice at the IATEFL conference next week.]

Personalization and open platforms: putting teachers first
There are a lot of things that technology can do for teachers, and for students (although, like with any toolset, technology can be misused).  That said, perhaps the single most important benefit that technology will bring to ELT is personalization. As we saw in Part 1, good teachers already provide adaptive content, an adaptive classroom, and an adaptive curriculum to personalize learning. But well-managed technology can make this teacher-driven personalization exponentially more effective and more powerful. 

We all know that it’s needed. Today no one doubts the profound limitations of the traditional top-down, global, one-size-fits-all coursebook. Less discussed, but just as problematic, is the standard, global ELT methodology taught in CELTAs and the like; it’s just not reasonable to think that the same ELT methods and approaches will be optimal for a student in Brazil and at the same time be optimal for a student in Saudi Arabia. Personalization is the future of ELT, and technology holds the key … actually it holds two keys. One key is the algorithmic adaptive learning we looked at in Part 1, carried out appropriately and with an understanding of its limitations. The other key is open platforms.

By the term “platform”, I refer to web services for ELT, usually cloud-based, that combine learning management tools (LMSs) with course content.  Oxford, Macmillan, Pearson, and Cambridge all have various platforms. But these platforms are not open. They are closed platforms that allow limited teacher options and are basically another medium to deliver our old friend: the traditional, top-down, global, one-size-fits-all coursebook. And this is another current limitation of algorithmic adaptive learning for ELT: currently it is being used only with closed platforms for major publishers.

By the modifier “open, I refer not necessarily to free or open source, but rather to platforms that are designed to be teacher-driven, not content-driven. Open platforms empower teachers and schools to engage with and re-configure content to fit the specific needs of their students, culture, context, customers, etc. Open platforms allow teachers to use a range of content from a range of publishers, curate this content, re-configure and re-order the content, and, using the same authoring tools, create their own content and slot it in wherever appropriate. Open platforms are course creation and delivery tools; the resulting material can be delivered in class (printed out or by way an IWB or projector), online, or, optimally, as part of a blended learning course. Moodle, Sakai, Canvas are examples of an open platforms, as is my project, English360. By providing publisher content as the raw material, as well as authoring tools for teachers to create content, open platforms allowing teachers to manage that material as they see fit, and course design moves from a top-down, publisher-only product to a collaborative process among publishers, teachers, and the students themselves.

Again, teachers have been assembling bits of content into courses for ages.  If you walk back into the teachers’ room in many language schools, you’ll find teachers lined up at the photocopier. As I’ve explained to many publishers over the last few years, most of these teachers are not photocopying material because they want to steal. They are photocopying because they want to personalize. They may take a page or two from one coursebook, then some from another, then another page from a grammar resource, then copy an article from a local magazine with the exercises and tasks they made that exploits that authentic content. They select this content because they understand the students’ needs and interests, their strengths and weaknesses, their cultural background, the course rhythm…all the critical data that only a teacher can know, and that publishers can only vaguely anticipate and generalize in order to produce global coursebooks.

Closed platforms are a publisher delivery mechanism for these global, generic coursebooks. Open platforms are tools that empower teachers by allowing them to do digitally what they do with a photocopier, by personalizing content and providing differentiated instruction that adapts to the needs of each student, each class, each school, each culture.  Teachers are the ones working in the classroom, talking to parents, and experiencing each student’s learning. That role is indispensible to any realistic approach to adaptive learning in ELT, and open platforms make that possible.

Open platforms and adaptive learning
For adaptive content, homework or pre-class work on an open platform can be created on the fly with a few clicks, based on the knowledge of the teacher of the profile of each student or each class. Adaptive course creation on open platforms can range from out of the box publisher content, to curated courses taking units, pages, or specific tasks from a range of coursebooks and resources, together with course content authored by teachers specially for local needs. All can be easily adapted to learner performance by teachers on an ongoing basis as the course progresses. And for larger scale projects, a set of these customized courses, together with assessment tools, can evolve over time by way of the teacher community of practice, again based on a developing understanding of learner requirements.  As stated above, at IATEFL next week, my colleague Valentina Dodge and I will be presenting a case study of a program with an adaptive curriculum that we set up for seven colleges in the Middle East, using an open platform driven by the local teacher community of practice.

What about algorithmic adaptive learning hooked up with teacher-driven open platforms?  Absolutely. In a blended learning program, this would be the best of both worlds: letting a machine do what it can do best, and letting humans do what humans do best.  Using Knewton’s data infrastructure for the subset of content that is done online and is objectively scored will be a very powerful tool and of real value to every teacher.  What’s interesting is that in most blended learning courses, the online component is often language presentation, focused practice and form-focused drilling, and pre-teaching vocab or other exercises that can be automatically graded. So that would be appropriate to the micro-adaptation that Knewton’s infrastructure can provide.  The results could potentially feed in to classroom and curricular adaptation, although as noted before this will be more difficult with language teaching as opposed to topics such as science or maths. That said, the value of open platforms in blended learning is not really about what happens online, it’s about what happens in the classroom when the class can focus 100% on what only humans do: communicative, interactive activities, task-based and project-based learning, and messy, open and subjective communication.

So, let’s look again at the three categories of adaptation (content, classrooms, and curriculum) and contextualize each of these within the primary purpose of adaptive learning, personalization.

adaptive taxonomy

So, let’s sum up both posts and look at where we go from here. Adaptive learning is something that teachers have been doing forever, manually, by way of adaptive content, adaptive classrooms, and adaptive curriculum.  It was always so natural for us that we didn’t call it anything special, other than “good teaching”.  Recently, mechanical, algorithmic adaptive learning has become possible, for a very specific subset of learning activities: those that are discrete, online, and objectively scored. In ELT this is largely “micro-adaptive” content, such as form-focused practice and drilling, delivered as online homework or as part of a blended program.

Given the holistic nature of language and language teaching, algorithmic adaptive learning will be harder to implement and easier to misuse than in other domains, such as science or maths.  Led by Knewton, it has gained considerable commercial momentum, and, as “the next big thing”, major ELT publishers have all jumped on the bandwagon.  It is these very same publishers that give us generic, one-size-fits-all coursebooks that many teachers believe focus on commercial success rather than learning outcomes. These coursebooks are certainly the antithesis of personalization, which has come to be seen as a critical road forward for ELT, and, when delivered digitally, come to us by way of closed platforms that severely limit the role of the teacher. So there is concern.

But, technology has brought us other tools, and one of these is the open platform. Open platforms are designed to support teacher-driven personalized learning, and do so by way of adaptive content, adaptive classrooms, and the adaptive curriculum. For the right subset of content delivered through an open platform, algorithmic adaptive learning can be an invaluable tool, and the insights gained through the resulting big data can help teacher adapt the classroom and curriculum to student needs. But language learning is more subjective and holistic than, say, maths, and so for ELT professionals, Knewton’s toolset is both more limited (to a subset of content types) and more easily misused (because the big data insights come from a limited data set).

In short, algorithmic adaptive learning is a potentially powerful tool that need not be feared by teachers, but instead looked at judiciously and used appropriately. But adaptive learning is not done only by machines: teacher-driven adaptive learning through open platforms can empower teachers to be the protagonists of ELT technology, rather than its victims.
Featured Photo Credit: Billy Wilson Photography via Compfight cc. Text added by eltjam.


  1. Enjoyed the articles, as well as the insightful comments. What most LMSs are missing today is quite simple–fun, feedback and friends! Learning needs to be fun for our digital natives. It needs to provide tailored feedback; lastly, it needs to allow students to connect.

    1. @Christina and Cleve

      I just posted on my blog about having fun and creativity.

      I think I would add independence to fun. There are quite a few ‘private social networks’ out in the www and I think one or two have been successful. But, my view is that education needs to be independently connected and social. Opt in or out, and be able to move around. A sense of community is essential but also independence. Because, if the student leaves one institution for another – they need to be able to take their community with them. From this I can see global niche communities emerging that share and discuss what is important to them – and crossing cultures for the greater good. Mmm Facebook is ‘and all in option’. It´s either in or out. And I don´t like that. Twitter less so – zoom around with more privacy. But, I would like to see an app like ‘whatsapp’ designed for education, entrepreneurship, and community. As well as teachers. It may well be that this will be the beginning of students direct relationship with teachers – and that schools and government are left out of the equation. Which I think is a healthy thing. Independence means taking responsibility for learning and we are not really doing that with kids today.

      At least that´s how I see it.

      1. Dear Cleve,

        In all the talk of adaptive learning, the people at Knewton seem to believe that they can help us become incrementally better at learning some things (and I accept this as true). But by the same token they never talk about helping us become exponentially better at language learning.

        One reason I think this is true is that they view language learning from the lens of discrete items and nothing is a better example of “discrete” than grammar. Despite attempts to change the way we look at language teaching, the mainstream history of language teaching over the last 50 years is the history of grammar’s ebb and flow.

        As a “conversation” teacher I often must resist the call from my students to look at grammar as the core constituent of language learning (largely because of tests). Now I have a new reason to fear grammar. In the past it was because I felt it often got in the way of the acquisition process. IMHO, grammar, because it is easy to measure, puts accuracy above fluency WAY TO EARLY. IMHO, grammar instruction also attempts to subtly separate grammar knowledge from word knowledge WAY TO EARLY. But now I have a new reason to fear grammar. Now I will fear grammar because it will become VERY profitable for publishers to promote self-study grammar instruction using Knewton-like systems (see your thoughts on self-study).

        It is reasonable to conclude that the longer a publisher can keep a student slogging thru the details of a grammar online the more money they can make (esp. if we are right in our assessment that much new grammar based instruction can eliminate the teacher). In the future, it would seem that the more “they” can make grammar the core of language learning, the more they have to gain FINANCIALLY. This is a scary thought because I am afraid that what is sound financially will also drive pedagogy and testing.

        My question, in a very leading way to you is this: Do you think it is possible that these incremental advances in grammar instruction that we see on the horizon from systems like Knewton, twined with a greater need for “efficacy” in publishing, and spurred on by the ability to potentially make more money from self-study grammar learning tools will result in a greater focus on grammar (and discrete item) study at the beginning levels of language learning?

        Are large publishers again, as always, favoring incremental changes over exponentially larger changes? And because Knewton-like systems are not direct change agents themselves (they view themselves in service to publishers) aren’t these machine learning systems also just agents of small, incremental change?

        By the way, I don’t believe I have an axe to grind. I am just trying to understand the implications for the teacher of how the people who have a the biggest stake in this process (you said billions of dollars) will attempt to drive machine learning systems keeping in mind that Knewton views publishers, not teachers, as their strategic clients.

    2. HI Christina

      Agree completely that they need to be more fun and more social. I haven’t really found much of a solution to that, alas. In my experience, it seems most of the fun in a blended learning course is in the classroom, and it’s more fun than usual because most of the individual, heads-down activities have been done previously online.

      Also, I recently came across some to-be-published research showing that for online activities the students actually preferred the traditional drills to the social component such as forums or Facebook activities! Now, that may well be because, in this specific case, those social activities were poorly designed. But many students said something along the lines of “when I’m on Facebook it shouldn’t be for English…”

      It’s an interesting problem, and like anything else in ELT, it’s tough to generalize about. But, fun is good, so as you say, we need to keep pressing on….

      1. Thanks, Cleve! Will you post that research when it comes out? I think many would find it interesting. Our belief up to now seems to be that LMSs need to be more like social networks.

  2. Robert,

    I’m not a member of the ELT Jam team. But I know their contact details are up somewhere on the site.

More comments

Leave a Reply

Your email address will not be published. Required fields are marked *


Other related posts

See all

New decade, new Jam!

My English learning experience – 6 lessons from a millennial learner

Deconstructing the Duolingo English Test (DET)