This document provides an introduction to deep learning for natural language processing (NLP) over 50 minutes. It begins with a brief introduction to NLP and deep learning, then discusses traditional NLP techniques like one-hot encoding and clustering-based representations. Next, it covers how deep learning addresses limitations of traditional methods through representation learning, learning from unlabeled data, and modeling language recursively. Several examples of neural networks for NLP tasks are presented like image captioning, sentiment analysis, and character-based language models. The document concludes with discussing word embeddings, document representations, and the future of deep learning for NLP.