PALABRIA: Language Learning Enhancement through Linguistic Rule-Based Analytics and Artificial Intelligence

In the academic world, and in the professional world, the use of language is essential to solve different tasks, as well as to communicate the results. However, it is very common for speakers to make errors of expression (even in their native language) such as, for example, incorrect use of grammar, or poor choice of register. In order to assist such communicators in the use of written language, artificial intelligence (AI), including the recently popular generative AI, can become an ally in identifying norm-sanctioned usages, and providing appropriate and personalized feedback.

Although AI solutions in education exist for different subjects, the application in linguistics has been less explored and requires additional complexity, since by its nature it is a field with usually more open rules. Moreover, existing solutions that address linguistic aspects are limited by often developing natural language processing systems without taking into account linguistic rules or having limited support for feedback or the use of text-based analytics.

Thus, the main objective of this project is to improve students’ written language proficiency and usage by combining AI techniques (machine learning, generative AI, etc.) and linguistic rules. To achieve this, the project will propose: 1) a research framework including scenarios enriched with AI elements and linguistic rules, 2) the development of improved system solutions for error detection and feedback on language use, using AI and learning analytics, 3) a technological and linguistic framework, with the definition of an architecture that overcomes current limitations, and 4) pilot experiences to analyze and validate the proposed solutions and to know the most appropriate AI techniques and their degree of accuracy in this context.

• Partners: Universidad Carlos III de Madrid (coordinator).
• Duration: 2024-2026
• Project id: PALABRIA-CM-UC3M