Emilia: a tool to facilitate learning French in school
By Conseil de l’innovation du Québec | Published on September 24, 2024
Problem and objectives
Problem
- The workload required for the correction of written productions means that secondary school French teachers have a much heavier task than others.
- The time required to correct the texts generates a delay in providing feedback to the student which varies between three and six weeks.
- The complexity of the task, the lack of practice and the long delays before obtaining feedback make it very difficult for the student to really engage, or even improve.
Objectives
- Emilia is a tool that greatly reduces the time spent by French teachers on correcting texts. This frees up valuable hours that they can use to plan more stimulating lessons for their students.
Proposed solution
Solution
Emilia pre-corrects the students' texts. The role of the teacher thus shifts from correction mode to validation mode, which makes her work much more interesting and more focused on feedback on the ideas put forward by the student. Emilia also allows the collection of numerous data on the skills deployed by the student. This data can then be used to help each student discover their personal strengths and challenges.
Principle
The tool fully supports fault detection as well as judgment of ideas. He corrects the texts in a neutral and rapid manner while precisely following the requirements according to the progression of MELS learning. Production is therefore evaluated according to the constraints of the level. However, the teacher always remains in control, he can modify everything.
User experience
The reduction from approximately 16 minutes to less than 4 per student copy makes the experience terribly attractive. With Emilia, a high school teacher will take around 6 hours to validate 100 copies, which could previously take more than 35 hours.
Role of AI
The AI will correct the text while respecting the author's intentions, systematically coding each type of error. It will also be able to detect alerts in texts when the tone is too intense, for example if a student expresses hateful remarks or suicidal thoughts. These copies will be identified among the others a few minutes after submission.
Impacts
Main outcomes
Increased motivation: teachers are delighted to finally have a tool designed primarily for them. The fact that feedback is rapid engages students more in their learning.
New educational software on the market: no solution as complete and rigorous exists for the high level of complexity required for correcting texts according to ministerial expectations.
Intelligent automation: the use of Emilia by more than 30 teachers leads us to validate our tool. By iteration loop, we improve the coding of errors according to user requests.
Conditions for success
Team mobilised
In education, having a team completely dedicated to such a solution was essential to the success of the project. This is the bet that Collège Sainte-Anne has made by dedicating resources solely to research and development of the Emilia tool.
A large project team has been set up. It notably included two consultation tables totaling 12 primary and secondary school teachers. French teachers revised the correction of the validation texts and a doctoral lecturer from UQAM was also involved.
In the research and development of phase 2, several classes were contacted to test learning environments. They were thus able to give their advice regarding the representation of the results of analysis of writing skills.
The project manager works directly in the establishments, which greatly facilitates obtaining feedback from students and teachers.
Collaborations
The project was carried out with technology partners MoovAI and NexApp.
Funding received
The project received financial support from the Sainte-Anne Foundation, the INVEST-AI program (managed by IVADO Labs) and private donors.
Project stages
We will complete version 1 in spring 2025. At that time, dynamic progressions will be implemented between the different productions, which will highlight the progress made. We will also have completed the linkage with the progression of learning, thus allowing all stakeholders to better target their interventions in order to promote essential learning, level after level. This version will then be available to other stakeholders in the Quebec education sector (school service centers, private schools). Eventually, regional versions will be developed.
Project stages - to come
The development of version 2 will be launched in 2025. It will target learning environments to amplify and accelerate educational processes. Our research and development in this area (1350 hours) has allowed us to identify six environments that appear to have a much greater impact than is currently recognized by the evidence. Everything is already functional and tested; all that will remain is to add these modules for young people.
A version intended for parents is also in the plans. All we need is a larger production team.
Context of the solution
Customers
Emilia is intended for teachers, students and soon parents. We are primarily targeting Quebec, but our solution could also be used elsewhere in the French-speaking world. Ultimately, it could even be adapted to other languages and countries. Whether we teach here or elsewhere, the challenge of correcting written productions is the same: it is a time-consuming task that harms learning due to infrequent feedback loops. This is the problem that Emilia wants to solve.