Grounded Language Modeling for Automatic Speech Recognition of Sports Video