The article compares how well a machine learning model called Recurrent Neural Network (RNN) can analyze people's feelings in Bengali and English text. They collected a dataset of over 17,000 texts from different sources like social media. The RNN model was able to accurately analyze sentiment in both raw and translated text, achieving 82.9% accuracy for Bengali text and 84.5% accuracy for English text.