Leben in Deutschland: An Intelligent Tutoring Game for the German Citizenship Test

Abstract

To apply for German citizenship, each candidate must complete a multiple-choice questionnaire of 33 questions with at least 17 correct answers. The official question database contains a total of 310 questions covering various topics such as history, politics, and values. While there are already courses, websites and apps to prepare for this test, there exists no digital serious game that creates an engaging environment and supports the learners individually. The aim of this proposed intelligent tutoring game “Leben in Deutschland” is to support learners in preparing for the test, by situating them in a German town, tracking question mastery in repeated trials, providing feedback on failed questions and allowing them to pose open questions on the topics. Feedback and answers to open questions are given using the Large Language Model (LLM) llama3.2-3B with Retrieval Augmented Generation (RAG). The background knowledge contains encyclopedias by the Bundeszentrale für politische Bildung (bpb) and public broadcast video-transcripts. We conducted a mixed-methods pilot with 4 German speaking participants (age 18 - 45) with 3 of them having a migration background and one of them having completed the citizenship test already. The participants rated the usability as 4.4±0.6 and the perceived improvement over learning without the game as 4.9±0.3 on a 5-point Likert scale showing overall high satisfaction with the game. Three of the participants rated the LLM-generated feedback as helpful, one noted it lacked depth. In qualitative feedback, the participants liked the simple yet effective gameplay loop and proposed additional features such as quests that integrate the learning content in the interaction with NPCs. These results indicate that an LLM‑augmented serious game can deliver engaging, individualized preparation for the German citizenship test. In future work we want to evaluate the system in a trial with a cohort that is preparing for the citizenship test and explore multilingual extensions.