Minority languages are increasingly entering digital environments through institutional websites, educational platforms, automatic translation tools, corpora and AI-mediated communication. This visibility may appear, at first sight, as a form of protection. However, the digital recognition of a minority language often requires selection, normalization and standardization: one orthography, one grammar, one “representative” variety, one searchable and machine-readable form. The very technologies that promise inclusion may therefore risk reducing the internal plurality of minority languages, especially when oral traditions, local variants and intergenerational forms of transmission are poorly represented in digital data. This paper proposes a semiotic analysis of minority language protection in the age of artificial intelligence, using Italy as a case study within a broader international question. Italy offers a particularly rich field of observation because it combines legal recognition of historical linguistic minorities with complex local realities, where language is not only a communicative code but also a cultural memory, a marker of belonging and a symbolic resource for communities. The central research question is: can AI support linguistic justice without transforming living minority languages into simplified and standardized objects? The theoretical framework draws on semiotics of culture, particularly the concepts of semiosphere, centre and periphery, cultural memory and translation between codes. From this perspective, minority languages are not treated merely as “small languages” requiring technical preservation, but as semiotic environments in which communities negotiate continuity, identity and difference. Standardization is interpreted as a cultural operation that legitimizes some forms while marginalizing others. AI mediation intensifies this process because machine learning systems usually privilege written, stable and institutionally validated data. The paper focuses especially on communication across generations. Older speakers often preserve situated, oral and locally embedded linguistic practices, while younger generations increasingly encounter minority languages through digital interfaces, social media, school materials and AI tools. This shift creates both risks and opportunities. On the one hand, AI may reproduce linguistic hierarchies by favoring dominant languages and standardized varieties. On the other hand, if developed with community participation, transparent datasets and sensitivity to variation, AI may become a tool for documentation, education and intergenerational dialogue. The paper introduces the concept of semiotic linguistic justice to describe an approach in which language protection is not limited to legal recognition or digital visibility, but also includes plurality, memory, variation and community agency. The contribution aims to connect sociolinguistics, language policy, digital humanities and human–AI communication, showing that the future of multilingual dialogue depends not only on how many languages AI can process, but on how respectfully it can engage with their internal diversity.
Renato Ongania (Sun,) studied this question.