This study presents a language- and context-independent framework for emotion recognition in social media content using a large-scale dataset of English and Turkish Twitter posts. Leveraging Ekman’s six basic emotions and neutrality, the research applies distant supervision and domain adaptation techniques to automatically annotate and classify emotions using both traditional machine learning and advanced deep learning models, including BERT and its variants. The proposed framework enables cross-cultural analysis of emotional responses to brand communication, offering significant theoretical, methodological, and managerial implications in areas such as brand management, customer relations, and public opinion mining.