AI Writing Tutor for Lower Secondary Education: Development and Evaluation of an Adaptive Writing Tutor
An AI-powered tutor to support students in lower secondary education in their writing skills with personalized feedback.
Duration: January 2026 – July 2026
Status: Ongoing
Educational Level: Lower Secondary Level
Topic: Artificial Intelligence AI, Digital Skills & Literacy, Digital Tools
Keywords: Writing Skills, Adaptive Learning, Artificial Intelligence
Initial Situation
Writing proficiency in Swiss lower secondary education is under pressure, as shown by the 2022 PISA studies. At the same time, with class sizes of around 22 students, teachers cannot provide the individualized, iterative feedback that students’ writing development requires.
Objectives
Therefore, our goal is to address this gap by designing, developing, and evaluating a new AI writing assistant for Swiss lower secondary students using TexTutor. This AI assistant provides adaptive scaffolding and motivational prompts in the form of elaborative feedback, tailored to students’ developmental and learning needs. It is based on large language models that have been fine-tuned with texts written by Swiss students. The system incorporates age-appropriate guardrails and is aligned with Lehrplan 21 competencies.
Method
We will use a mixed-methods design with a design science research approach. First, we will conduct interviews with teachers and review literature on AI-supported writing instruction. Then, we will identify how AI writing tutors can support writing and determine which feedback functions are most useful. Additionally, we will conduct an empirical study in lower secondary education in Switzerland, incorporating TexTutor into regular writing units. We will collect quantitative and qualitative data to derive initial design requirements and answer our research question.
Planned Translation
From the beginning, teachers and students are co-designers of our application. We integrate TexTutor into actual writing assignments aligned with Lehrplan 21 competencies in Swiss schools. To ensure the outcomes are immediately usable, we will provide a teacher-ready implementation guide, a reusable constructive-alignment framework, a set of rubrics with Swiss-German examples for lower secondary education, and a dashboard concept with relevant practical indicators.
The project will provide evaluated design principles for developmentally appropriate, elaborative feedback in AI writing assistants for adolescents in Switzerland. To ensure our project’s relevance, we will conduct an empirical study in Sek I classes using TexTutor during regular writing units. Our goal is to collect quantitative data based on self-efficacy and learning analytics (e.g., feedback uptake and revision cycles). Additionally, we will collect qualitative data to capture perceptions, usability, and contextual factors that influence adoption.
