Robots can help teachers and schoolchildren learn the new Latin-based alphabet

The Kazakh language transition from Cyrillic to the Latin alphabet in Kazakhstan raises challenges in teaching the whole population to write and read in the new script. The NU CRP “CoWriting Kazakh” project (2020-2022) proposes a unique interdisciplinary approach by integrating innovative solutions from robotics, computer vision fields, and pedagogical strategies from education, linguistics, and cognitive sciences that will assist diverse demographic groups in this challenging endeavor.

The research is led by Dr. Anara Sandygulova, NU School of Engineering and Digital Sciences, in collaboration with Dr. Anna CohenMiller from the NU Graduate School of Education, Dr. Wafa Johal from the University of New South Wales, Dr. Thibault Asselborn, and Dr. Pierre Dillenbourg from École Polytechnique Fédérale de Lausanne, and Dr. Danna Summers from Baishev Aktobe University.

As the first milestone of the project, the team implemented autonomous behavior for a  robot that would assist and motivate primary school children in learning the new Kazakh script and its associated handwriting. The system integrates AI-driven recognition of children’s handwriting in Cyrillic in order to convert it to Latin letters using a digital tablet with a stylus and a humanoid robot. A demonstration of this system received the “Best Demo” award at the 2020 ACM/IEEE Human-Robot Interaction conference.

 As Dr. Sandygulova explains:

“A human-robot interaction application is designed to help primary school children learn a new script and also its handwriting system. The system is inspired by the paradigm from learning sciences known as Learning by Teaching where children become a robot’s teacher of the Kazakh language. It engages them into the so-called “Protégé effect” as they commit to the interaction with the robot and unconsciously feel that they are responsible for its success. The robot plays a unique social role here as it makes children practice their Latin script skills without even noticing it. Thus, the system is more attractive to children in comparison to a traditional teacher or tablet-only approaches.”

 The robotic system was deployed in a series of experiments with primary school children in order to investigate a few hypotheses.

 We firstly investigated what script (Cyrillic vs Latin) children should use to teach a robot to write Kazakh words. Results demonstrated that girls learned more letters when they directly taught the robot using Latin spelling of Kazakh words, while boys learned significantly more when spelling the words using Cyrillic and then observed the robot’s correct spelling of the Latin-based Kazakh words. These findings suggest that the system should adapt to individual learner’s strengths and weaknesses. This work was published in the proceedings of the 2020 ACM/IEEE Human-Robot Interaction conference”, explained Dr. Sandygulova.

 The project involves mentoring of junior researchers including multiple research assistants. Zhanel Zhexenova, Nazarbayev University Research Assistant, continued on to explain other results of another experiment: “In an attempt to understand whether the robot brings any effect in comparison to other teaching methods, our results suggest that children gained similar knowledge of a new script in all three teaching approaches, however children’s likeability ratings and positive mood change scores demonstrate significant benefits favoring the robot over a traditional teacher or tablet-only approach.”  These findings can be seen in published form in Frontiers in Robotics and AI journal.

 An integral part of the research highlights inclusive education and awareness to the needs of all students. Dr. Sandygulova explains that “Apart from motivating and assisting Kazakh children, can we estimate the impact of this political decision on children who have learning disabilities such as dysgraphia or dyslexia? We need to understand whether introducing a new script would cause children to have negative side-effects for their handwriting and reading skills. To understand that, we analyzed the transfer of handwriting skills from Cyrillic to Latin script of 200 Kazakh children. We examined dynamic data obtained from digital tablets’ stylus, such as tilt, pressure, altitude and other dynamic features responsible for diagnosing dysgraphia. The results demonstrate that Kazakh primary school children were successfully able to transfer their fine motor skills from Cyrillic to Latin. These findings are forthcoming in the npj Science of Learning journal.”

 As Dr. CohenMiller explains, “This research provides essential insight to issues of equity and inclusion in education. We hope to address some challenges that people could face during the alphabet change such as across socio-economic backgrounds, geographic region, ethnicity, age, and gender. In order to effectively teach and learn the Latin-based Kazakh alphabet, a central aim is to develop various online and offline innovative pedagogical tools based upon differentiated learning strategies relevant to various learning scenarios and individuals.” 

 Future steps include adaptation of the CoWriting Kazakh system for remote and online learning as a web and chatbot applications, additional integration of pedagogical practices and equity to suit various learning scenarios and individuals, such as older adults and teachers, and a longitudinal evaluation by the end-users.

To learn more about the project, see the following  video for an explanation: