LangMentor is an AI language tutor that chooses topics, lessons and exercises based on your level and progress. Its decisions draw on a connected set of learning matrices linking the language’s content, the depth of practice and your learning history.
How the learning matrices work together
The aim is to map the language as a whole and connect that map to an individual learning path.
- The concept matrix maps language concepts, topics and their relationships. It defines what there is to study.
- The depth-of-practice matrix — Practice Map sets out how to work on each concept: recognising a form, choosing it, transforming it or building a sentence. For verbs, this includes conjugation across tenses and persons. Progress against the matrix shows what you have practised and what needs more work.
- The word-frequency matrix structures vocabulary practice from common words to less frequent ones. As you remember the words you have studied, the tutor introduces less common vocabulary. The goal is to help you start speaking sooner and build a path towards full command of the language.
Your progress is linked to this structure: the concepts you have studied, how far you have worked through them and your results. The tutor uses that history to choose the next step: a new topic, more practice or a review.
What happens in a lesson
The tutor explains a topic, sets exercises and checks your answers. You can ask for a hint or a further explanation along the way.
At the end of the lesson, the results and rating are saved in the database. They inform the choice of future lessons and exercises. Each concept has its own review schedule.
Language in the context of a country
For Spanish, the exercise generator includes a collection of information about Spain’s culture and everyday life. It provides context for language practice, with the aim of helping learners get to know the country as they study its language.
My role and the current version
I am responsible for the architecture, learning logic and code. The tutor is in development, and my friends and I use the current version to learn.
Comments can be attached to actions during a lesson, together with the sequence that led to the system’s answer. I use this context to investigate feedback and decide what to improve next.