My research focuses on natural language processing, with a focus on developing trustworthy and generalizable language AI systems. These systems aim to understand human concepts, speak human languages, and perform logical reasoning. Additionally, they are designed to be robust, reliable, and adaptable to new domains, languages, and modalities. Currently, my research includes several key directions: (1) Robustness of large langauge models (LLMs), (2) Multimodal concept learning and alignment, (3) Multilingual text understanding, and (4) Data-driven knowledge extraction.