Compréhension en image

When we set up this project, we had a dual objective: - to get our students involved in a practical, meaningful teaching project that would help them to develop their professional skills - to ...

When we set up this project, we had a dual objective: - to get our students involved in a practical, meaningful teaching project that would help them to develop their professional skills - to help French language learners improve their comprehension by creating mental images. Using the tool we have created, we are offering exercises to improve the comprehension of poor readers. The aim is to create a mental picture of the story told by the text. At a more basic level, it means forming a mental image of a scene from a sentence. We wanted to produce a sufficient volume of sentences and images to train the pupils, and not just to assess them. Our ambition was to produce 1,200 sentences and 1,200 images in the form of multiple-choice questions sorted by level of complexity in a database. The final tool was put together on a web page enabling teachers, or anyone else who needed it, to easily generate a PDF document according to their preferences, free of charge. The tool allows them to generate a PDF according to the level and number of exercises required. A correction is also provided at the end of the PDF.

This project brings together artists, educators, linguists and computer scientists to design a bank of sentences and images as well as software to generate exercises. The activities were carried out as follows 1) From March to April 2023: Creation of 1,000 sentences on 10 levels of difficulty by the teachers and the linguist in collaboration with the two Belgian schools. A number of tests were carried out in the classroom (middle school/high school). 2) 15 to 17 May 2023: a transnational trip to the Brassart school in Lyon, where students illustrated and corrected almost 500 illustrations. 3) June to October 2023: testing of the image sentences in order to refine the design patterns. 4) 13 to 16 November: second design session, this time at the Montpellier site. The students produced almost 1,000 semantically rich illustrations. In the end, 872 of them were selected. Alongside the teaching and artistic work, adults undergoing retraining worked with their Reactor trainer to design software to generate the images and sentences used to produce the exercises. The format of each image had to be standardised to optimise their size for the web.

The final tool is 'comprendre.netlify.app', a web portal that can be accessed by anyone free of charge and which enables PDFs of exercises to be generated at will on 10 levels of difficulty with a simple click. Each question consists of a sentence and 4 images (multiple choice). The learner must then choose the image that best matches the sentence. At level 1, the distinctions are simple, for example between images of coffee cups, tea cups, a broken cup and Grandpa's cup. Level 10 includes complex sentences linked by a conjunction such as "because" or "but", for example with a squirrel that nibbles because it is hungry, another that only nibbles when it is hungry and a third that is full but does not nibble. During the sentence development stage, we also had to standardise the signs and arrows used in the illustrations to make them easier to understand. These exercises were very well received by the French teachers in the two partner schools. Initial feedback from our network has also been promising. We have not carried out a formal evaluation of this project in terms of the pupils' academic results.