Abstract:

Context:
Natural language learning belongs to the ill-defined domain where the development of Intelligent Tutoring Systems (ITS) made little progress. This is due to the fact that these learning environments are founded on the problem-solving and that is not a typically task for ill-defined domain. In this work we devise an ITS to learn multi-tonal words of the Mooré language.
Most languages spoken in Africa belong to the class of tone languages. A tonal language has polysemous words depending on the level tonal of its pronunciation. A wrong pronunciation can lead to misunderstanding. Mooré is the most widely spoken language in Burkina Faso and in some of its neighboring countries such as Ivory Coast , Ghana and Togo. Before the pandemic covid-19, there were learning centers where Non-Governmental Organizations and embassy workers and medical staff usually went to learn Mooré. Nowadays only few of these centers are re-opened.
These workers learn this language to communicate with the local population who mostly do not understand French. Even though, French is the official language, it is not spoken by the majority of the population, generally in the countryside.

Goal:
We aim to build an ITS to learn the Mooré. This learning environment provides transcription activities with goal to distinguish multi-tonal words. As the target public of learners is literate in french, we developed the ITS those users.

Contributions:
With two Mooré experts, an educational consultant and a linguistics lecturer, we identified the relevant learning tasks to integrate into the system. The main parts of this work are as follows. First, we used knowledge engineering methodology to identify and describe the knowledge and processes of the domain and tutoring modules. We obtained a specification that is suitable for other tone languages. Then, we represented the domain knowledge. The model tracing has been used for the transcription rules. As the space of wrong transcriptions is not too large we preferred the buggy rules to match student errors in order to provide remedial feedback. Finally, we studied the system feasibility. We used Petri net formalism to present the system. The Petri nets model allows us to prove the consistency of the system and other requirements of its functionality. Thus, we made sure there is no deadlock state in the system.

Keywords:

Intelligent Tutoring System, Tone Language.