Adapting STEM Automated Assessment System to Enhance Language Skills

Enhancing language skills is a duty for many institutions, schools and universities; Information and Communications Technology is nowadays supporting language teaching, heading also to the integration of foreigners. According to previous studies, the use of a Technology Enhanced Learning Environment can foster language competences through online tests. The goal of this paper is to show and discuss different typologies of automatically evaluated questions useful to learn languages and designed with an Automatic Assessment System, already successfully used for teaching STEM disciplines: Science, Technology, Engineering and Mathematics. This work has been carried out thanks to the collaboration between the Department of Mathematics and the Department of Foreign Languages. Five different typologies of questions are analysed: "Combining sentence elements" asks to combine given elements to form at least three correct and meaningful sentences; "Find the mistake" asks whether a sentence is grammatically correct and to explain why; "The Target puzzle" asks to join words together using precise links; "Scrambled text" asks the student the correct sequence of sentences in a scrambled text; "The Encrypted Crossword puzzle" is a crossword puzzle without definition clues. Linguists consider these question types effective to develop language competences and have already used them profitably in traditional teaching and language learning. To engage the learner and increase motivation, we will develop an implementation where the questions are self-evaluated, always available, with immediate and interactive feedback. These tools could be used in online language courses and in all those teaching activities that are carried out remotely by the student alone. All questions are outlined with the explanation of the grading code and a triplet of descriptors; they describe the Performance, Requirements and Objectives of each question and allow further studies on automatic detection of the relatedness between different language learning objects, in order to construct adaptive language tests.