Second language acquisition (SLA) in English can be revolutionized by artificial intelligence (AI) because it can provide personalized learning methods for learning English. Although AI has made advancements in this area, the standard SLA systems that are familiar with often fail to provide real-time, personalized feedback that can enhance language proficiency. This gap needs to be addressed with a more adaptive, data-driven approach. This study presents the Intelligent Galactic swarm optimized feedforward neural network (IGSO_FNN) method to improve SLA. Integrating GSO with feedforward neural networks (FNN) solves the issue of learning weights and model parameters, increasing accuracy and personalizing the language learning experience. IGSO significantly improves the model's adaptability, particularly in model training when dealing with complex linguistic data. The dataset contains text and speech data of vocabulary usage, pronunciation accuracy, and grammatical patterns from varying proficiency levels. Experimental results demonstrate that the IGSO-FNN approach outperforms the traditional method, with learners benefiting from increased accuracy (97%), precision (95%), F1 score (93%), recall (96%), and MSE (0.04). The IGSO-FNN method is a scalable, adaptive solution to enhance SLA with real-time feedback and personalized learning paths.