Socioculturally Aware Language Technologies

Despite recent advancements in natural language processing (NLP), current models often overlook the critical influence of social context on language. This thesis develops methods to incorporate sociocultural context into NLP models, enhancing their performance, fairness, and generalizability. By focusing on cultural, community, and personal contexts, this thesis aims to enrich NLP models with a deeper understanding of the social dynamics that shape language. The thesis introduces novel approaches for measuring and operationalizing social context, as well as for integrating this information into model architectures and datasets. Through empirical studies across various NLP tasks, this work demonstrates the effectiveness of the proposed methods in improving model performance and addressing social biases. Ultimately, this research contributes to the development of more robust, inclusive, and human-centered language technologies.