Computational Methods for Integrating Vision and Language