Periodic Reporting for period 2 - iRead (Infrastructure and integrated tools for personalized learning of reading skill)
Reporting period: 2018-07-01 to 2019-12-31
Globally, UNESCO reports that there are almost 800 million people who are illiterate. This is not just a problem for developing countries. In England, for example, 25% of young adults have poor literacy compared with an average of only 9% in the top performing countries in Europe (Kuczera et al 2016). Illiteracy can result from antecedents such as social opportunity including language and literacy practices at home, efficacy of pedagogical support and the child’s reading difficulties (Rose, 2006). Within compulsory education, about 1 in 10 children are struggling readers who are unable to master decoding as it is traditionally taught, as a result gradually being left behind in the classroom despite their potential. There is increasing evidence to suggest that technology can be used to foster literacy and to our current interest, reading.
Two barriers exist to fully realising the potential of these learning applications are: (1) the lack of personalisation that can meet the varied needs of each student and (2) the lack of an integrated approach that exploits the pedagogical opportunities of each individual tool to enhance different reading processes. The ambition of iRead is to:
• Develop a flexible, scalable and cost-effective cloud-based software infrastructure which supports personalised learning technologies;
• Take a holistic approach to reading that considers both reading comprehension and word decoding skills as equally critical parts of the reading process;
• Upgrade learning applications previously shown to foster reading development into personalized technologies by interfacing with the iRead infrastructure;
• Orchestrate the teaching process through learning analytics tools that allow on-going formative assessment and other teacher resources supporting the use of the technology by teachers;
• Adopt a wide reaching approach targeting primary school children acquiring reading and writing skills (i) either making typical progress in learning to read (ii) children who struggle with reading due to dyslexia and (iii) children who are learning a foreign language.
We then defined seven ‘domain models’ shaping the project’s learning goals, and developed dictionary content resources used by the project applications. These represented the needs of novice readers in four languages (Greek, English, German, Spanish), struggling readers (English, Greek) and children learning English as a foreign language and languages. Domain models capture the language features that have to be mastered when learning to read, they are described through difficulty levels and prerequisites which determine progression. Also, we have developed specialized annotated dictionaries which are linked to the domain models form the primary source of educational content. Using the domain models, the following four applications have been designed.
A text classification tool: efforts were directed towards personalising text selection for learners based on their reading needs as well as reading goals. In the process of developing text analysis tools, we also had to develop syntax analysis tools for all the languages covered by the project; a product that can stand on its own and prove valuable in the wider context of Natural Language Processing.
A literacy game and interactive e-books: we developed a games application called Navigo designed to teach reading. Through sixteen puzzle game mechanics, our game engages learners across a wide breadth of learning objectives; covering both word and sentence level reading content. Through iterative learner-centred design, involving feedback and evaluation from students and teachers, the game has been refined to address pedagogical, engagement and usability concerns resulting in the game being available on the Google Play store.
A Reader app: the Reader Amigo has been designed to scaffold children’s reading through personalised embedded supports and other instructional features. The app design includes design feedback from project partners and Primary school teachers as well as technical feedback from project partner Dolphin to assess feasibility. A Beta version of the app has been delivered, evaluated for usability by students at several Primary schools and launched on Google Play store. Project partners have sourced accessible child appropriate content to use as part of the evaluation phase of the project in schools.
The iRead project partners have designed, implemented and deployed the cloud software infrastructure which supports the Navigo game and the Amigo Reader in offering a personalized “learning to read” experience. The infrastructure consists of several components handling an array of sophisticated tasks including user and domain modelling, user management (authentication, authorization), resource management (eBooks, specialized dictionaries), data logging (recording user actions during game-play and reading), syntax analysis, text annotation, and feeding Navigo with learning content.
Since Spring 2019, we have been conducting evaluations of the Navigo and Amigo apps at schools. In total, 3,900 students in primary Education across the UK, Spain, Greece, Sweden, Romania and Germany have participated to date in researcher-led studies, or are currently using the apps as part of their classroom teaching. Alongside the school evaluations, the project game, Navigo, is available for download by the general public through an open pilot targeted to parents.
• The integration of four language domain models and related resources, applied to modelling of learning of three diverse learner groups for a number of applications (e.g. Navigo games, Amigo Reader)
• Game activities supporting three stages of reading fluency for a range of language levels
• Machine learning based methods for syntax analysis for four languages
• Demonstrating the feasibility of supporting personalized interactive eBooks.
• Creating a collection of user-actions during gameplay that can be utilized in research in fields of education and linguistics.
• More effective personalised teaching practices that scale across different educational contexts
• Improving reading motivation and skills of primary school children, and in turn lifelong learning and social opportunity
• Teacher training to support the delivery of digital approaches to education
• Raising public awareness on the importance of reading in learning and the opportunities of digital learning