In speech recognition, the language model is used to compute or assign the probability of a word sequence w from an input acoustic sequence. it is typically probabilistic and depends on previously derived grammers, as well as the application domain and the speaker.
