This thesis investigates the design, development, and evaluation of Jointales, an AI-assisted interactive storytelling system aimed at supporting collaborative narrative creation between neurodivergent (ND) and neurotypical (NT) children. Using a Design-Based Research (DBR) methodology, the study follows an iterative cycle, from conceptual framing to prototyping, user testing, refinement, and expert evaluation, to explore how digital storytelling environments can be made inclusive, cognitively accessible, and emotionally safe for diverse neurocognitive profiles. The prototype integrates Large Language Model-generated narrative guidance, a dual-mode interaction structure (Simplified Mode for ND children and Full Mode for NT children), cooperative challenge tasks, and accessibility-centered design features such as segmented text, Read Aloud support, reduced sensory load, and micro-feedback animations. Quantitative data were collected using the System Usability Scale (SUS), User Experience Questionnaire-Short Version (UEQ-S), and a custom-designed Jointales Evaluation Form, while qualitative insights were gathered through observations, think-aloud protocols, and expert/child feedback. Findings demonstrate that the refinement cycle significantly enhanced the system’s performance: SUS scores increased from 74.38 (Good) to 86.88 (Best Imaginable), and UEQ-S results reached the Excellent category for both pragmatic and hedonic quality. Qualitative analyses highlight that sensory safety, readability, and predictability are essential for ND children, while interaction richness, movement, and choice-based exploration are crucial for NT children. Experts confirmed that the system supports social-emotional behaviors such as turn-taking, helping, and cooperative problem-solving. This study contributes a set of inclusive design principles for AI-supported interactive storytelling, demonstrating how neurodiversity-informed design can shape future child-AI interaction frameworks. Findings suggest that adaptive narrative systems, when grounded in accessibility and co-experience, hold strong potential for fostering communication, empathy, and shared imagination among diverse child users.
Questa tesi analizza la progettazione, lo sviluppo e la valutazione di Jointales, un sistema di narrazione interattiva assistito dall’intelligenza artificiale, progettato per supportare la creazione collaborativa di storie tra bambini neurodivergenti (ND) e neurotipici (NT). Attraverso una metodologia di Design-Based Research (DBR), lo studio segue un ciclo iterativo, dalla definizione concettuale alla prototipazione, dai test con gli utenti alla fase di refinement e alla valutazione esperta, con l’obiettivo di esplorare come gli ambienti digitali di storytelling possano diventare inclusivi, cognitivamente accessibili e emotivamente sicuri per profili neurocognitivi diversi. Il prototipo integra una guida narrativa generata da Large Language Models, una struttura di interazione a doppia modalità (Simplified Mode per bambini ND e Full Mode per bambini NT), compiti cooperativi e funzionalità di design orientate all’accessibilità, come testi segmentati, supporto alla lettura ad alta voce (Read Aloud), riduzione del carico sensoriale e micro-animazioni di feedback. I dati quantitativi sono stati raccolti attraverso la System Usability Scale (SUS), l’User Experience Questionnaire-Short Version (UEQ-S) e un Jointales Evaluation Form appositamente sviluppato; parallelamente, intuizioni qualitative sono state ottenute tramite osservazioni, protocolli di think-aloud e feedback di esperti e bambini. I risultati dimostrano che il ciclo di refinement ha migliorato significativamente le prestazioni del sistema: i punteggi SUS sono aumentati da 74.38 (“Good”) a 86.88 (“Best Imaginable”), mentre i risultati UEQ-S hanno raggiunto la categoria “Excellent” sia per la qualità pragmatica sia per quella edonica. Le analisi qualitative mostrano che sicurezza sensoriale, leggibilità e prevedibilità sono fondamentali per i bambini ND, mentre ricchezza interattiva, movimento ed esplorazione basata sulle scelte risultano cruciali per i bambini NT. Gli esperti hanno inoltre confermato che il sistema favorisce comportamenti socio-emotivi quali il rispettare i turni, aiutare l’altro e risolvere problemi in modo collaborativo. Questo studio propone un insieme di principi di design inclusivo per sistemi di storytelling interattivo supportati dall’IA, dimostrando come un approccio informato sulla neurodiversità possa contribuire alla definizione dei futuri modelli di interazione bambino-IA. I risultati suggeriscono che sistemi narrativi adattivi, fondati su accessibilità e coesperienza, possiedono un forte potenziale nel promuovere comunicazione, empatia e immaginazione condivisa tra bambini con profili neurocognitivi differenti.
Jointales: empowering neurotypical and neurodivergent children's communication and collaboration skills through LLM-driven interactive storytelling
Kaptan, Elif Ekin
2024/2025
Abstract
This thesis investigates the design, development, and evaluation of Jointales, an AI-assisted interactive storytelling system aimed at supporting collaborative narrative creation between neurodivergent (ND) and neurotypical (NT) children. Using a Design-Based Research (DBR) methodology, the study follows an iterative cycle, from conceptual framing to prototyping, user testing, refinement, and expert evaluation, to explore how digital storytelling environments can be made inclusive, cognitively accessible, and emotionally safe for diverse neurocognitive profiles. The prototype integrates Large Language Model-generated narrative guidance, a dual-mode interaction structure (Simplified Mode for ND children and Full Mode for NT children), cooperative challenge tasks, and accessibility-centered design features such as segmented text, Read Aloud support, reduced sensory load, and micro-feedback animations. Quantitative data were collected using the System Usability Scale (SUS), User Experience Questionnaire-Short Version (UEQ-S), and a custom-designed Jointales Evaluation Form, while qualitative insights were gathered through observations, think-aloud protocols, and expert/child feedback. Findings demonstrate that the refinement cycle significantly enhanced the system’s performance: SUS scores increased from 74.38 (Good) to 86.88 (Best Imaginable), and UEQ-S results reached the Excellent category for both pragmatic and hedonic quality. Qualitative analyses highlight that sensory safety, readability, and predictability are essential for ND children, while interaction richness, movement, and choice-based exploration are crucial for NT children. Experts confirmed that the system supports social-emotional behaviors such as turn-taking, helping, and cooperative problem-solving. This study contributes a set of inclusive design principles for AI-supported interactive storytelling, demonstrating how neurodiversity-informed design can shape future child-AI interaction frameworks. Findings suggest that adaptive narrative systems, when grounded in accessibility and co-experience, hold strong potential for fostering communication, empathy, and shared imagination among diverse child users.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Kaptan.pdf
solo utenti autorizzati a partire dal 19/11/2026
Descrizione: Research Thesis
Dimensione
73.89 MB
Formato
Adobe PDF
|
73.89 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/247272