Enabling Sardinian Language Learning via Word Translation and Speech Synthesis based on Artificial Intelligence

Abstract

The rapid advancement of digital technologies has reshaped language education, creating opportunities for innovative teaching practices that can be extended to minority and endangered languages. Sardinian, officially identified as one of Italy’s historical minority languages, presents a highly complex linguistic landscape due to its internal variations, uneven geographical distribution, and fragile vitality. Such factors led to significantly weakened intergenerational transmission, primarily due to the language’s marginalization within formal schooling and the scarcity of digital learning resources. From this perspective, integrating advanced linguistic-technological tools into the classroom represents a strategic action of cultural preservation and revitalization, enabling both the safeguarding and promotion of an intangible linguistic heritage and a higher engagement of learners with their local identity. In such a context, text-to-speech technologies, initially developed for accessibility and assistive purposes, are being increasingly valued as pedagogical tools for fostering linguistic and phonological competence. Building on this potential, this work addresses the limited intergenerational transmission and the lack of modern teaching resources for Sardinian by introducing a digital educational environment based on a generative text-to-speech model. This initiative is expected to enhance learners’ phonological and metalinguistic skills while supporting the revitalization and digital vitality of the Sardinian language.