My research investigates how language and thought co‑evolve in technologically mediated environments.
I work at the intersection of philosophy of language, cognitive science, linguistics, AI‑augmented learning, and digital literacies.
My long‑term goal is to develop a unified cognitive framework for understanding human learning in the age of AI.
My work is guided by a central question:
How do language and thought co‑develop when learners interact with AI‑mediated environments?
To answer this, I combine:
Exploring the cognitive mechanisms linking linguistic structures and conceptual representations.
This includes:
Studying how AI tools reshape learning processes.
Focus areas:
A pedagogical experiment conducted with 60+ learners in a Cameroonian ZEP classroom tested a novel method: learners wrote descriptive texts about literary characters, then used AI image generation to visualize them. The method operationalizes the Short Loop of the Cognitive Trapeze Model — where the AI acts as a visual feedback mechanism for the learner’s written production. This approach directly supports the concept of embodied knowledge and proves that AI can enhance descriptive writing even in low-resource, offline-first contexts.
👉 Explore the AI Writing Lab: AI Writing Lab
In collaboration with the IA4ZEP program, I am designing a mobile application that uses Automatic Speech Recognition (ASR) and Large Language Models (LLM) to support guided reading sessions in priority education zones.
The application strictly follows the official 5-step guided reading methodology (situation, reading, grid, confrontation, synthesis) prescribed by the Cameroonian curriculum for 6th grade French classes. It provides:
This tool bridges the gap between linguistic decoding and deep semantic comprehension, directly validating the Cognitive Trapeze Model and aligning with national educational priorities.
The production of AI-generated comics for guided reading directly operationalizes the concept of embodied knowledge from the Cognitive Trapeze Model. The comic serves as a visual and narrative anchor that supports semantic comprehension, particularly for neurodiverse learners. This resource was tested with 60 learners in a ZEP classroom, demonstrating the feasibility of multimodal AI-generated pedagogy in low-resource contexts.
👉 See the educational resources: IA_4_ZEP Resources
A comprehensive ASR evaluation across 13+ texts and 26 learners in a ZEP classroom revealed a striking pattern: 4 learners (15.4% of the sample) exhibited acoustic and attentional signatures consistent with neurodivergent profiles (dyspraxia, dysarthria, ADHD spectrum). Using Whisper Tiny, Medium, and Large as diagnostic proxies, we applied four evidence‑based criteria:
These patterns are not errors—they are acoustic signatures of neurodevelopmental differences. The remaining 22 learners (84.6%) showed typical progression curves aligned with grade level and reading confidence.
Pedagogical implications for Cognitive Trapeze:
These findings validate the dual‑layer ASR+LLM architecture and confirm that Cognitive Trapeze’s adaptive design is not an optimization—it is a necessity for inclusive, neurodiversity‑aware pedagogy.
Investigating how learners in African contexts engage with digital tools.
Includes:
Developing computational simulations to model:
My work combines:
This methodological diversity bridges theory, empirical data, and computational simulation.
Based in Ngaoundéré (Cameroon) and collaborating across Europe and Africa, my research builds bridges between:
This strategy strengthens both scientific and societal impact.
My MSCA project Cognitive Trapeze integrates all four research axes.
It provides a coherent framework for:
The project is built on four core hypotheses aligned with the work packages and empirical predictions: