technology for the language teacher
Adaptive learning
Philip Kerr
In this series, we explore technology-related themes and topics. The series
Downloaded from [Link] by gaupeguy on 25 March 2026
aims to discuss and demystify what may be new areas for some readers and to
consider their relevance to English language teachers.
‘Adaptive learning’, in the context of language learning and teaching,
means different things to different people. In the most general terms, it
can be defined as a way of delivering learning materials online, in which
the learner’s interaction with previous content determines (at least in part)
the nature of materials delivered subsequently. The process is automated,
dynamic, and interactive. Its purpose is to generate a personalized
learning experience.
Personalization ‘Personalization’, however, is as slippery a term as ‘adaptive learning’
in educational contexts. Although there is widespread agreement
that ‘learning in the future ought to be more personalized’ (Duncan
2013), there is no generally agreed single definition of what this means
(UNESCO Institute for Information Technologies in Education 2012). The
disambiguation by the US Department of Education Office of Educational
Technology (2010: 11) is helpful as a way of understanding different kinds
of adaptive learning:
Individualization Differentiation Personalization
Learning goals are the Learning goals are the The learning objectives
same for all students, but same for all students, and content as well as
students can progress but the method or the method and pace
through the material at approach of instruction may all vary.
different speeds. varies according to the
preferences of each
student or what research
has found works best for
students like them.
Adaptive learning is an educational technology, rather than a method. It
can be used in the service of any of the three approaches described above.
In English language learning until now, there have been more uses of
adaptive learning for individualized approaches than for differentiated
or personalized ones. It may prove to be the case that the affordances of
ELT Journal Volume 70/1 January 2016; doi:10.1093/elt/ccv055 88
© The Author 2015. Published by Oxford University Press; all rights reserved.
Advance Access publication October 7, 2015
adaptive learning lend themselves more to the former than to the latter,
but the current rarity of differentiated or personalized adaptive learning is
more likely a reflection of the relative rarity of such courses in traditional
face-to-face contexts.
Flashcards and The simplest form of online language instruction that can be called
related apps adaptive is a digital flashcard app (usually for vocabulary) which
incorporates automated spaced repetition of the target language item.
Algorithms determine the order, frequency, and number of times in
which a learning target (typically a word) is presented, and this is based
on the individual learner’s response to their previous exposure, via the
app, to this item. The particular set of target items can be determined by
a teacher (or by the program itself ), or by importing sets of words that
Downloaded from [Link] by gaupeguy on 25 March 2026
are associated with other learning material (for example a coursebook),
or by individual learners themselves. Programs of this kind are becoming
increasingly popular and sophisticated. Recent developments include
elements of gamification (for example badges, progress bars, leader
boards), game-like practice of target items, the linking of target vocabulary
items to authentic online texts (both written and video), and interactive
flashcards (for example learners are presented with gapped sentences)
with hints and automated intelligent feedback.
More complete language learning programs (usually centred around a
combination of vocabulary and sentence-level grammar, but also offering
some skills development) include, or soon will include, adaptive elements.
One such learning program, Duolingo (2013) describes its approach as
follows:
Every time you finish a Duolingo lesson, translation, test, or practice
session, you provide valuable data about what you know and what you’re
struggling with. Our system uses this info to plan future lessons and
select translation tasks specifically for your skills and needs.
Another uses adaptive learning in its review lessons. These are scheduled
‘at optimal intervals [and] designed to help you efficiently retain what you
learn without wasting time reviewing material that you already know well’
(Rosetta Stone n.d.).
Although most apps like these allow learners some freedom in selecting
the particular language items they want to learn (for example extra
vocabulary items), the overall learning goals (for example reaching a
previously defined ‘level’) remain fixed. The approach is individualized
with some, usually minor, features of differentiation.
Big data and Individualized learning programs adapt to a relatively small amount
learning analytics of data generated by the learners. Essentially, this is the learner’s
responses to a given learning task. In order to offer a more differentiated
approach, it is necessary to collect substantially greater amounts of data.
This includes personal information (such as age, sex, home address,
academic and course attendance records, disciplinary records) which
can be obtained when the educational platforms that deliver the adaptive
courses are integrated with administrative systems. It also includes
the sort of information that is routinely collected for the purposes of
Adaptive learning 89
business analytics (for example to analyse and predict buying patterns for
companies like Amazon)—such as page views, the length of time spent
on a page, and click rates—not only when interacting with the learning
material but also during other online activity. For this data to be valuable,
i.e. for it to lead to actionable insights, it needs to be aggregated with
that of other learners, and the greater the number of learners, the more
valuable it is considered to be. These numbers can now be very large,
as publishers strike deals with large educational institutions or acquire
chains of language schools themselves. (In 2013, for example, Pearson
bought the Brazilian chain of franchised language schools, Grupo Multi,
with 800,000 students (Mance 2013).)
The analysis of such big data in education, it is claimed, makes it
Downloaded from [Link] by gaupeguy on 25 March 2026
possible to recommend or modify personalized learning paths, to ‘create
productive peer groups, and free up class time for creativity and problem
solving’ (Ferreira 2013). Although the use of such big data and learning
analytics is widespread in the United States in higher education, it has yet
to make any documented impact in English language teaching. Contracts
have been signed between Knewton, one of the largest companies to
offer this technology, and major ELT publishers (including Pearson,
Macmillan, and Cambridge University Press), but efficacy research is still
largely unavailable in the public domain.
Granularity and Adaptive learning assumes that learning is linear and cumulative. It
learning outcomes also assumes that what is to be learnt can be analysed and defined at a
‘granular’ level (it can be broken down into small ‘learning objects’ or
‘learning chunks’) and that these can be organized into a hierarchical
‘knowledge graph’, one granular piece of learning is a prerequisite for
another. The acquisition of these learning objects needs to be measurable;
adaptive learning presupposes largely predetermined learning outcomes.
However, sophisticated adaptive learning systems are dynamic and allow
for some modification of knowledge graphs as a learner interacts with the
material.
Although these assumptions appear to be relatively unproblematic in
the learning of some school subjects, such as mathematics, the ‘mastery
learning’ approach is more contested in language learning. Larsen-
Freeman (1997: 151), for example, argues that ‘learning linguistic items is
not a linear process—learners do not master one item and then move on
to another. In fact, the learning curve for a single item is not linear either’.
Learning objects do not need to be linguistic items in adaptive language
learning (they could, for example, be defined in terms of skills or a
learner’s interaction with a text), but, in practice, they invariably are. That
being the case, adaptive learning is often associated with discrete-item
product-oriented syllabuses. Thornbury (2015) offers a succinct analysis of
the problems associated with such an approach.
Evaluation The Common European Framework of Reference for Languages
(Council of Europe 2001) provided an evaluation framework for English
language competence, consisting of a complex grid of skills, divided
into six levels. More recent developments have attempted to offer more
‘precise’ numerical descriptions of English language competence. The
90 Philip Kerr
Cambridge English Scale, for example, has numerical values from 80
to 230, and the Pearson Global Scale of English ranges from 10 to 90.
At the same time, there have been attempts to correlate linguistic items
(especially vocabulary and grammatical structures) with points or bands
on such frameworks. One example of this is the English Profile project,
which describes itself as ‘setting industry standards for English language
learners’ (see [Link]
The tendency for adaptive courses to see language ‘learning objects’
within hierarchical ‘knowledge graphs’ as discrete items of vocabulary and
grammar is reinforced by the ready availability of granular scales, such
as those described above. The learner’s developing language knowledge
(rather than competence in language use) is continuously evaluated
Downloaded from [Link] by gaupeguy on 25 March 2026
and recorded. By comparing large data sets of learning outcomes with
large data sets of learning variables, it is also possible to evaluate the
instructional materials themselves and the effectiveness of both courses
and teachers, when adaptive learning is a part of blended instruction. If
the desirability of such evaluation is accepted (and it is accepted by most
national governments and intergovernmental organizations, including the
OECD and the World Bank), ‘technology-based learning and assessment
systems [are seen to] be pivotal in improving student learning and
generating data that can be used to continuously improve the education
system at all levels’ (US Department of Education Office of Educational
Technology [Link].: ix).
Adaptive learning Adaptive learning has been described as a ‘hot concept’ that is ‘poised to
and the education reshape education’ (Webley 2013). Its appeal to educators comes primarily
industry from the promise of personalized learning. Its appeal to policymakers
comes from its promises of accountability, greater productivity, and
economy. Despite very high initial investment costs, online adaptive
instruction is cheaper than traditional alternatives such as face-to-face,
classroom teaching (Spring 2012: 2).
Many publishers see adaptive learning as a powerful response to the
business challenges they are facing. These include rampant pirating,
spiralling development costs, and the growth of open educational
resources. Adaptive learning services are sold on a subscription basis.
As an interactive service, they cost learners/customers more than
static (printed or digital) content and cannot easily be pirated. After
development, production and distribution costs are much lower, and sales,
marketing, and support services can be concentrated on large institutional
customers.
It is unsurprising to find that ELT publishers are deploying increased
resources to those parts of the world (South and Central America, the
Middle East, and East Asia) where there is a concentration of large
institutional customers and an interest in online courses with adaptive
elements. Lower-level (CEFR levels A1–B1) language learning materials
account for by far the highest volumes of sales in these markets. As we
have seen, adaptive learning appears to lend itself most readily to content
that is structured around discrete items of vocabulary and grammar.
At the lower levels, it is easier to map out such ‘knowledge graphs’ (by
Adaptive learning 91
analysing the frequency of the items), than it is at higher levels, where
there is likely to be more diversity of learning needs.
Big data, learning analytics, and adaptive learning have already
transformed the educational landscape in the United States and other
countries. They have been promoted by massively funded educational
foundations (Ball 2012) and private business interests, as education
is recast as an ‘industry’. But in English language teaching, change
is proving to be rather slower. There is a clear, global shift away
from traditional print-based classrooms towards the use of learning
platforms and blended models of instruction. There are now materials
(for example Doff, Thaine, Puchta, Stranks, and Lewis-Jones 2015) that
offer a limited amount of differentiation in learning paths, and adaptive
Downloaded from [Link] by gaupeguy on 25 March 2026
testing is becoming more commonplace. Spaced repetition apps are
proving popular. However, truly personalized English language learning,
facilitated by adaptive learning, remains in the experimental stages.
Because of the affordances of adaptive learning for the presentation of
linear syllabuses of discrete items, there is a risk that this new technology
will become no more than a sophisticated version of the various teaching
machines offering ‘programmed instruction’ that were developed before
modern computing. Adaptive learning does not necessarily entail either
a linear syllabus or a behaviourist approach, but it may result in the
reinforcement of both within the world of ELT.
References Mance, H. 2013. ‘Pearson buys Brazilian language
Ball, S. J. 2012. Global Education Inc. Abingdon: school chain Grupo Multi’. Financial Times,
Routledge. 3 December. Available at [Link]
Council of Europe. 2001. Common European cms/s/0/3a9a601a-5bf2-11e3-931e-00144feabdc0.
Framework of Reference for Languages: Learning, html#axzz3ktpOyEyz (accessed on 5 September 2015).
Teaching, Assessment. Cambridge: Cambridge Rosetta Stone. n.d. ‘What is adaptive recall™?’ Article
University Press. on website Rosetta Stone. Available at [Link]
Doff, A., C. Thaine, H. Puchta, J. Stranks, and [Link]/temp/articles/en_US/Text/What-is-
P. Lewis-Jones. 2015. Empower. Cambridge: Adaptive-Recall (accessed on 5 September 2015).
Cambridge University Press. Spring, J. 2012. Education Networks. New York, NY:
Duncan, A. 2013. ‘Enabling the future of learning’. Post Routledge.
on blog White House, 17 December. Available at https:// Thornbury, S. 2015. ‘G is for Granularity’. Post on
[Link]/blog/2013/12/17/enabling-future- blog An A–Z of ELT, 3 May. Available at https://
learning (accessed on 5 September 2015). [Link]/2015/05/03/g-is-for-
Duolingo. 2013. ‘Duolingo’s data-driven approach granularity/ (accessed on 5 September 2015).
to education’. Post on blog Duolingo, 31 January. UNESCO Institute for Information Technologies in
Available at [Link] Education. 2012. ‘Personalized learning: a new ICT
post/41960192602/duolingos-data-driven-approach- enabled education approach’. Policy Brief March 2012.
to-education (accessed on 5 September 2015). Available at [Link]
Ferreira, J. 2013. ‘Big data in education: the 5 types en/files/[Link] (accessed on 5 September 2015).
that matter’. Post on blog The Knewton Blog, 18 July. US Department of Education Office of Educational
Available at [Link] Technology. 2010. National Education Technology
blog/ceo-jose-ferreira/big-data-in-education/ Plan 2010: Transforming American Education:
(accessed on 5 September 2015). Learning Powered by Technology. Alexandria, VA: US
Larsen-Freeman, D. 1997. ‘Chaos/complexity science Department of Education. Available at [Link]
and second language acquisition’. Applied Linguistics [Link]/sites/default/files/[Link] (accessed on
18/2: 141–65. 5 September 2015).
92 Philip Kerr
Webley, K. 2013. ‘The adaptive learning revolution’. in a number of coursebook series, including
Time Magazine, 6 June. Available at [Link] Straightforward and Inside Out. He is also the author
[Link]/2013/06/06/the-adaptive-learning- of Translation and Own-Language Activities and
revolution/ (accessed on 5 September 2015). two e-books, A Short Guide to Adaptive Learning
in English Language Teaching and How to Write
Vocabulary Presentations and Practice. He blogs
The author about technology and language teaching at https://
Philip Kerr is a teacher trainer and ELT materials [Link]/.
writer. He is the author or co-author of titles Email: philipjkerr@[Link]
Downloaded from [Link] by gaupeguy on 25 March 2026
Adaptive learning 93