Children’s interest in reading has been in steady decline for the past twenty years. This is shown in several studies, such as the international PIRLS study of reading achievements and attitudes to reading among fourth grade students. There is now an ongoing interdisciplinary design project aimed at finding new ways to increase interest in reading. Together with the Education Office in Norrköping, researchers at LiU are investigating whether, and if so, how, artificial intelligence (AI) can increase interest in reading with the help of a reading robot.
The robot used is a social robot, i.e. a human-like interactive robot. It was purchased from an external company but students and teachers at schools in Norrköping have been asked to come up with suggestions for what the robot should be able to do.
“If you develop technology to be used in the classroom of the future, you have to do it together with those who are there,” says Susanne Severinsson, docent in pedagogic practice and one of the research leaders.
Students contribute ideas
Susanne Severinsson, associate professor in pedagogic practice with a focus on special education at the Department of Behavioural Sciences and Learning. Photo credit Anna Nilsen To find out what the children and teachers want a reading robot to be able to do, the researchers have held workshops with them. The students were also asked to imagine what the robot would look like and then draw or build robots using LEGO blocks. Based on this, the researchers have come up with a number of possible design concepts.
“We could develop a robot that functions as a librarian who provides suggestions for literature based on the student’s interests and reading level. Or a reading coach who gives assignments, coaches the students and can explain difficult words. But what we tested as a first step is a robot that can hold a conversation about a book that the student has read,” says the other research leader Mattias Arvola, senior associate professor of cognitive science with a particular interest in interaction design and user experience.
The robot leads the conversation
To investigate how such a book conversation might turn out, the researchers have conducted group observations of students interacting with the robot. The response from the children was positive.
“They enjoyed talking to the robot and wanted to see it again,” says Emma Mainza Chilufya. She is a PhD student in cognitive science and controlled the robot during the book talks.
The robot is about eleven years old and speaks Swedish at the same level as children of that age. This is to make the children feel comfortable talking to the robot and not have to think about their replies. Sometimes the robot’s questions led to discussion between the students about the plot of the book, but usually they waited for new questions from the robot. It became clear that the conversation had to be driven by the robot.
“The students were happy to participate in the book talks, however, and they showed great interest in what the robot said,” says Susanne Severinsson.
“Many of them were curious about how the robot worked and how it could change skin colour. They asked if it could speak English and if it could change hats. Some of them wanted a female voice and asked if it could change, so I demonstrated a few different voices,” says Emma Mainza Chilufya.
There were many questions, but not everyone was impressed. Some students found the robot boring as it lacked arms and could not move. Others felt it was scary and that it was staring.
The reading robot, Emma Mainza Chilufya, PhD student in cognitive science, and Mattias Arvola, senior associate professor of cognitive science. Photo credit Anna Nilsen
The robot as a companion or character
The teachers selected four books that the robot could talk about and for which the researchers wrote the script. Two thicker books for students who read often and well and two easy-to-read books for those who find reading more difficult. During the talks, the robot was usually a companion, a kind of reading pal, with whom the children discussed the book, but on some occasions the researchers tried having the robot be a character from one of the books.
“That was interesting, the students liked the character. Some even asked if he helped write the book. I just wrote that I didn’t know. Writing short, relevant answers was the biggest challenge I encountered,” says Emma Mainza Chilufya.
The Wizard of Oz as an evaluation method
The method the researchers used is called the Wizard of Oz method, after the book/film of the same name, where the wizard controls a character behind the scenes. In other words, they have simulated the interaction between the robot and the students by remotely controlling it based on pre-programmed behaviours and scripts.
“This is a good way to find out what kind of system is required for a meaningful conversation and how the technology needs to be designed to work in this particular usage situation. Once you find out, it will be easier to design and construct the robot in the best possible way,” says Mattias Arvola.
They are now in the data collection phase and many questions still remain. How will the interaction be done? What should the robot be able to talk about? In what ways?
“From my perspective on special education, we have to consider whether the robot should be aimed at the students who still have a long way to go when it comes to reading or at those who find reading easy. Or perhaps everyone. This is something we need to discuss,” says Susanne Severinsson.
Isn’t there a risk that students will be more interested in digital technology than in reading?
“Well, many children were curious about the technology and how the robot worked. But if they talk about a book at the same time, they can also discover that reading books is fun. And this is a step on the way,” says Mattias Arvola.
The method the researchers used is called the Wizard of Oz method, after the book/film of the same name, where the wizard controls a character behind the scenes. Photo credit Anna Nilsen
Facts about "Interest in reading - a design project about reading with a reading robot"
What: A three-year design research project in collaboration with the Education Office in Norrköping to investigate whether and how artificial intelligence (AI) can increase interest in reading among four different fourth grade classes in two schools in Norrköping, through interaction with a social robot.
When: The project runs between 2023 and 2025.
Who: The researchers involved in the project are Susanne Severinsson, associate professor in pedagogic practice at the Department of Behavioural Sciences and Learning; Mattias Arvola, senior associate professor of cognitive science at the Department of Computer and Information Science; Linnéa Stenliden, senior associate professor of pedagogic practice at the Department of Behavioural Sciences and Learning; Emma Mainza Chilufya, PhD student in cognitive science at the Department of Computer and Information Science; Tom Ziemke, professor of cognitive systems at the Department of Computer and Information Science and Anna Martín Bylund, associate professor in Pedagogic Practice at the Department of Behavioural Sciences and Learning.
Funding: The project is funded partly by the Norrköping Research and Development Fund and partly by the Swedish Research Council. The total funding is nearly SEK 6 million.