Öztürk, Ilcin; Muller, Cindy; Assmann, Dennis; Petrov, Leonid
Supervisor:Prof. Gudrun Klinker
Advisor:Plecher, David (@ne23mux)
Submission Date:30.03.2022 / 30.09.2022


Augmented reality and serious games are still being evaluated as a useful combination for interactive and digital game-based environments. DragonTale supports this idea and attempts to teach topics playfully without causing frustration or loss of motivation. The 3D serious game for learning the Japanese language has been in development since 2015. In the game, players must complete story-centric tasks, learn different kanji and kana, and face challenges to prove their knowledge. A previous big rework of the game added a battle system that offers engaging content to help memorize kanji through repetition. In this practical course, an AR version of the battle system as well as an AR sudoku puzzle were implemented. This allowed to conduct a user study to analyze the effects of game features such as the story, the controls, the battle system and the AR features on the learning success of \DragonTale. The results indicate that the integration of AR elements is a motivational factor for learning. However, the AR input methods used in this iteration of the game were unfamiliar to the participants and thus the group playing the AR version of the game showed worse learning outcomes. Nevertheless, aspects such as the story, design and battle system were positively received and results in long-term studies could give more insight into the effect of AR on language learning in serious games.

Augmented Reality

After the big revision of Dragontale, the AR elements in the game had to be reworked as well. Both the battle system and one of the mini-games, a sudoku puzzle, were chosen as good candidates to integrate AR features. Plane finding is used to get a real-world surface to augment the battle field or sudoku grid onto. Different approaches were chosen to convert the 2D input methods used previously into 3D input methods that make use of the AR environment.

Battle System - Choice2D Buttons3D Cube with Buttons on each side
Battle System - TextKeyboardPanels with Letters (spread out in space)
Sudoku2D ButtonsAR Markers

User Study

Over the course of two practical courses, two iterations of the same user study. The research questions for those studies were as follows:

  • Does AR alter how different game elements like story, battles, or controls, affect the learning success of a serious game like DragonTale?

  • Does AR improve learning outcomes when used in a serious game like DragonTale?

The participants in both surveys were assigned to two groups. One group played a non-AR version of the game, while the other group played the AR version. The survey used consisted of the NASA TLX and additional custom questions. The NASA TLX covered four different tasks - installation of the game, controls, fights and story. The custom questions covered general information and Japanese knowledge questions.

The first iteration of the study showed that the participants generally enjoyed the game and that the non-AR group performed a lot better in the Japanese knowledge questions. One possible explanation is that the AR input methods used in the game were unfamiliar to the users and distracted them from the learning content in the game. However, there were also big differences between both groups in regard to general procedure and play time as COVID-19 restrictions made the conduction of an in-person user study difficult. Therefore, it was decided to conduct a second user study once restrictions had further eased.

For the second iteration of the study, great care was taken to ensure that the procedure for both groups was the same. All participants were university students. They received an iPad with the game already installed and were allowed to play for 20 minutes before filling out the survey. Since the study was in-person, the conductors were able to answer questions more easily and participants did not have to spend time on troubleshooting. In this second iteration of the study, the non-AR group still performed a bit better in the knowledge questions, but the gap between the two groups was much smaller. Thanks to bug fixes done based on the feedback from the first study, the TLX scores of most tasks also improved.

More details to the studies and exact numbers can be found in the paper.


The study results show that the AR input methods chosen in this iteration of the game were distracting for the players and lead to worse learning outcomes. However, only the immediate learning outcome after a single play session was taken into consideration. This raises the question, if participants could adapt to these new AR input methods over the course of several play sessions. Since the groups playing the AR version of the game gave more positive feedback in general, the possible advantages of including AR elements in serious games might only become apparent in long-term studies.