Every Voice Matters: Scaling Dialectology with Computational Methods

New technologies are transforming how science is conducted: and dialectology is no exception. This discipline studies how languages vary across places, exploring differences in pronunciation, vocabulary, and grammar to understand how varieties evolve over time.

Traditionally, the study of dialects relied on pen-and-paper interviews. Today, internet-based citizen science is opening exciting new possibilities. One example is the Spanish Dialects App, a project that uses gamification to collect large-scale dialect and sociolinguistic data.

Building on this success, I aim to steer into Computational Dialectology, a new interdisciplinary field combining dialectology, sociolinguistics, register studies, and computational tools. This broader, more holistic approach allows us to study language variation across different factors, such as geography, gender, etc.

This work is urgent: dialects are disappearing and replaced by broader regional forms known as regiolects. At the same time, Artificial Intelligence and most Natural Language Processing systems have a strong bias towards standard varieties, sidelining under-resourced ones. This linguistic and cultural bias is silencing entire communities, even from the most widely spoken languages. By embracing linguistic inclusivity, Computational Dialectology offers a way forward harnessing new technologies to capture the full richness of language diversity and ensure that every speaker, no matter their background, is heard.