Artificial Intelligence - a tool for inclusivity in education?

Artificial intelligence (AI) is changing the very fabric of the world as we know it.  Some argue that AI poses an existential threat to humanity in the long term. Despite such fears, professionals in all fields are embracing AI to provide hitherto unimaginable insights to help them in their work.  

There are ethical debates about AI and inclusivity which humanity as a whole, and we as educators, must be mindful of; however, AI-assisted online resources in education are already here.  In this blog post, I'll examine how AI can address specific needs regarding reading skills and enable teaching strategies to foster an inclusive learning environment and positive experience for English language learners.

Students know better than anyone the crucially important role reading plays in language development from their experience in learning their native language. As they progressed through the educational system in their native language, they were gradually introduced to more and more complex texts and expanded their vocabulary year by year.  WIth regard to English, estimating how many words this amounts to depends on how words are counted, and whether the words are only recognized or can be used productively.  Based on recent research by Brysbaert et al (2016), a ballpark estimate is that the average American child has a productive knowledge of around 3,000 word families when they start school, and they acquire 1,000 word families a year on average until they graduate with a productive vocabulary knowledge of about 11,000 word families.

When it comes to learning a foreign language, students do not have the time to spend twelve years gradually building up their reading skills and vocabulary.  In fact, at our institution we expect students to progress from an A1 (Beginner) level knowing virtually no English to a B2 (Intermediate) level in only thirty-two weeks!!  In lexical terms, linguists have found that this means the students have to know as absolutely essential  the 3,000 most commonly used words in English which comprise about 85-90% of words in almost any English text.  On top of that, our students are going to use English in their university studies, so they also need to know absolutely essential common academic words, which make up 5-10% of academic texts.  Along the way of acquiring the roughly 4,000 absolutely essential common and academic words, they acquire about another 4,000-5,000 words from the 10,000 most commonly used words in English that are relevant to their specific interests.

To illustrate the importance of learning the 'right' vocabulary, look at the example below which shows a 'vocabulary profile' of the paragraph above illustrating the CEFR proficiency levels by colour-coding them according to the CEFR levels.  In this paragraph, only two words, "lexical, linguists", are not in the CEFR word lists.  

Looking at this same paragraph as an academic text, here is the 'vocabulary profile" of the ten academic word types

The main problem is that students have no systematic way to identify the level of their current lexical competence and then present a staged reading development programme to reach the goal of the minimum threshold they need to study at university in English.  It is impossible for this to be achieved with a lock-step curriculum and syllabus design that forces each student to progress through the same vocabulary learning process.  Moreover, classical reading curricula tend to focus on an 'average' student, putting students above or below this target at a disadvantage. Each student is unique and requires a reading programme that is tailored to their individual needs.  One approach is to harness the power of artificial intelligence and computer adaptive testing to foster a broader range of inclusivity.  In this blog post, I will review one such approach that has been provided to the teaching profession to use for free at

Readtheory allows a teacher to create a class to which their students enroll.  The reading level is assessed for each individual student, which the teacher can observe in the class dashboard and monitor the individual progress each makes.  At the same time, students can also see their own progress. 

The process is quite easy.  To get started, go to

Click SIGN UP to create a teacher's account, choosing the TEACHER role.

If you have a Google account, it is easiest to log in with your Google credentials, but you can create your own individual account if you wish.  Once you have an account, you can visit the DASHBOARD.  Click on the MY CLASSES icon (the teacher with the whiteboard icon) and you simply create a class by giving it an appropriate name.

Readtheory will assign a CLASS CODE to your class.  When students SIGN UP, they have to enter your class code, and they will automatically appear in your class roster.

Now is when the AI starts to assess each individual student's reading ability, which is a process that is ongoing and reassessed after each and every reading quiz.   Readtheory employs AI algorithms that apply 'computer adaptive testing' (CAT) methodology.  Before students are assigned readings, the algorithms determine their actual reading level.  It does this by requiring students to do eight reading tasks.  As students answer each reading task, their next reading task will be based on their success in the previous.  So, if the first reading task is easily answered by the student, the site will present a more difficult task.  If the student has difficulty answering the task, they will be presented with a lower level task the next time.  This is repeated eight times, at which time Readtheory can fairly accurately assess the student's reading ability.

You can see an example of one of my students here.  At first, reading texts at Grade 1 level were presented.  After the third quiz, the algorithms determined that level was too easy, so a Grade 2 level text was given to the student.  The student was successful at this level, so a Grade 3 level was chosen next.  This proved too difficult, so the next text was back down to a Grade 2 level.  The student continued to perform well at Grade 2 but had difficulty at Grade 3, so after the initial eight quizzes the student's reading level was set at Grade 2.  This process is the same for each and every student in the class.

Here is an example from the students who have started Readtheory in my class so far.  The names have been deliberately cut off when taking the screen shot.

As you can see from the Grade Level Performance graph in the links above, the initial Grade Level average of the thirteen students who participated was Grade 2.  However, each individual student has a unique profile that Readtheory will build on by monitoring their reading performance each and every time they do a reading quiz.

Here is another example of one of the students in my class, again illustrating how the algorithms continually adjust and challenge the students to develop.

You can select a time period to monitor students' progress, so it would be possible to monitor students on a daily, weekly or monthly basis.  KNOWLEDGE POINTS (KP) are awarded based in the number of questions answered correctly, as well as bonus points for completing a quiz without any errors. Somewhat like the game of golf, which has a handicapping system to allow players of all abilities to compete together, each student, regardless of their level, can earn KP on an even playing field.  If teachers are using gamification in their classroom, they can easily integrate the KP to promote an inclusive growth mindset for the reading sub skill.

Readtheory has been developed for American students going through K-12 education, so it is linked to the ELA Common Core Standards.  While not directly related to learning English as a foreign language, the standards are broadly compatible.  This framework allows students to clearly see their individual progress measured to an objective standard.  Based on the quizzes taken, students can see their mastery of the skills at the level they are reading which the AI continually tailors according to their individual needs and abilities as they evolve.

Readtheory is a good example of how AI can not only support language learning for the masses, but at the same time also provide an inclusive learning environment that is flexible to adapt to and reflect each individual student's needs and abilities as their specific language abilities evolve and develop.  This example of AI appears to be benign which the dedicated individuals who created it have provided to the wider community at no charge whatsover.  We must be aware of the potential for such AI to be less than benign, but if we enter this world with our eyes wide open, we can address many challenges associated with the desire to create inclusive learning opportunities, 

No votes yet

Research and insight

We have hundreds of case studies, research papers, publications and resource books written by researchers and experts in ELT from around the world. 

See our publications, research and insight

Sign up to our newsletters for teachers and teacher educators

We will process your data to send you our newsletter and updates based on your consent. You can unsubscribe at any time by clicking the "unsubscribe" link at the bottom of every email. Read our privacy policy for more information.